Skip to main content

Full text of "Survey of contemporary economics"

See other formats


Digitized by the Internet Archive 
in 2013 

A Survey or 


The participation of The American Eco- 
nomic Association in the presentation of 
this material consists in the appointment 
of a Committee on the Development of 
Economic Thinking and Information to 
plan a periodic review of economics, estab- 
lished in April, 1945, and of an appropria- 
tion of Association funds for this purpose. 

A Survey or 


Edited by 

University of California 


Published for 




Second Printing, March, 1949 

Third Printing, June, 1952 

Fourth Printing, February, 1954 

Fifth Printing, January, 1956 

Sixth Printing, November, 1957 

Seventh Printing, February, 1959 

Eighth Printing, April, 1961 

Ninth Printing, January, 1963 

Tenth Printing,- August, 1964 


<— SI — H— III— ■«- 


As readers of The Wealth of Nations know, the gain in product from 
the division of labor "is owing to three different circumstances; first, to 
the increase of dexterity in every particular workman; secondly, to the 
saving of the time which is commonly lost in passing from one species of 
work to another; and lastly, to the invention of a great number of ma- 
chines which facilitate and abridge labor . . ."* Economists have un- 
doubtedly increased their dexterity, saved time, and invented many 
machines by reason of the division of labor within their science. But they 
may also be uncomfortably aware of the picture drawn by Smith of the 
narrow specialist: "The torpor of his mind renders him, not only in- 
capable of relishing or bearing a part in any rational conversation, but of 
conceiving any generous, noble, or tender sentiment, and consequently 
of forming any just judgment concerning many even of the ordinary 
duties of private life. Of the great and extensive interests of his country 
he is altogether incapable of judging."" 

The American Economic Association would regard with dismay an 
incapacity of its members— because of a too confining division of labor- 
to conceive noble sentiments, judge of private duties, and assess the in- 
terests of their country. These qualities rest upon more than technical 
expertise: they require a knowledge of human history, philosophy, and 
politics. But, more immediately, they require that the economist look 
beyond the regression coefficients, propensities, balances of accounts, and 
equilibria of forces within his own laboratory to the whole arena of 
economic life. The primary purpose of the present volume is to provide 
to the economist outside a particular field an intelligible and reliable 
account of its main ideas— both analytical devices and their practical 
application to public policy— which have evolved during the last ten or 
fifteen years. For most of the less abstruse and technical subjects, it is 
hoped that the qualified layman, the beginning graduate student, and 
the public servant will also read with profit. 

x Adam Smith, The Wealth of Nations, Cannan ed., Bk. I, Ch. i; italics supplied. 
1 Ibid., p. 734. 


The period under review began in the worst depression in our history, 
continued through our biggest war, and extends into the time of our 
gravest concern for the future. It is little wonder that these years pro- 
duced an amazing efflorescence of economic ideas. Like their fellow 
citizens, economists were confronted with changes of awesome extent 
and rapidity and with the inadequacy of many of their older patterns of 
thought. Probably no decade or score of years in the entire history of 
economics could match those just passed, in innovation, sharpness of 
controversy, and volume of writing. 

To review the course of economic thinking in such a period and to 
appraise its results impose upon the writer of any of these surveys a task 
of formidable difficulty. It is unlikely that a qualified reviewer should not 
himself have taken an active part in recent discussions, and an active 
part is not generally quite neutral. And yet the expert is asked to with- 
draw himself from the fray and to take a detached view of his fellows 
and— still more inhuman— of himself. The reader of the essays which 
follow will not be surprised to discover that this feat has not always been 
achieved with equal success; but he may be disposed to treat this short- 
coming with indulgence: sometimes because of the difficulty of the un- 
dertaking, occasionally because of the brilliance of the performance. Only 
within limits does economics partake of the character of an exact science, 
and the judgment of the observer inevitably plays a role. Consequently, 
there is no official economics. The views expressed in this book do not 
necessarily represent those of The American Economic Association, the 
editor, the two critics associated with each essay, and possibly not even 
the author if he were to have the benefit of ten instead of two sets of 
proof. What, then, do they represent? 

These essays may fairly claim to present the considered thought and 
judgment of able scholars who are well aware of the trust involved in 
their delegation to the task. All of the authors have recognized their 
fallibility, and— so far as possible without the sacrifice of their own honest 
convictions— they have tried to incorporate or to meet the suggestions of 
critics and editor. In rare cases, the piece appeared from the mold without 
serious flaw at the first pouring. More frequently, the workman was con- 
tent only after a second, third, or even fourth attempt. Ideally the qualify- 
ing or dissenting remarks of the critics should also appear in print, to give 
fair warning to the reader that complete unanimity does not prevail. But 
the book is a stout volume as matters stand; and the process of arriving 
at a final version by the author, to appear with a final statement by critics, 
would stretch out toward infinity. These practical considerations deter- 
mined the actual procedure from the outset. If the book serves its purpose 


well, however, this will be attributable in large measure to the distin- 
guished economists who accepted the difficult and delicate role of critic, 
without having their voices heard by the audience, in order to further 
the success of the Association's undertaking. 

The present volume represents the fruition of extensive planning and 
preliminary work on the part of a Committee on the Development of 
Economic Thinking and Information, established by The American Eco- 
nomic Association in April 1945; and of an appropriation of Association 
funds for this purpose. Much of the arduous— and thankless— exploratory 
work was done by the chairman of the present committee, Professor J. J. 
Spengler, Duke University; by Dr. Eveline M. Burns, Columbia Uni- 
versity; by a former president of the Association, Professor A. B. Wolfe, 
The Ohio State University; by the faithful secretary of the Association, 
Professor James W. Bell, Northwestern University; and by the other 
members of the reconstituted Review of Economics Committee— Dr. 
Corwin D. Edwards, Dr. Paul T. Homan, editor of the American Eco- 
nomic Review, Professor W. Blair Stewart, and Dr. Aryness J. Wickens. 
The editor gives grateful acknowledgment of valuable advice and help at 
numerous junctures given by his colleague Professor William Fellner, 
and also by Professors Frank L. Kidner and Earl R. Rolph. 

Not all fields of economics worthy of inclusion in the present survey 
have actually been incorporated— population problems, social security, 
agricultural economics, etc.— to name only a few omitted. The somewhat 
arbitrary present selection of subjects, made by the Association's com- 
mittee, was dictated by the desire to achieve a portable volume; and the 
same consideration imposed upon the writers the citing of representative 
and outstanding works rather than the construction of exhaustive bibliog- 
raphies. The utilization of conventional fields in economics— overlapping 
and illogical as they may be in certain respects— seemed preferable, for 
ease of identification by the reader, to any newly decocted scheme. 
Finally, the choice of contributors and, in turn, their selection of subjects 
and substance answer to no grand design for a unified method or philos- 
ophy of economic thinking. The purpose of the book is not to impose an 
artful scheme upon the interpretation of recent analysis and policy, nor 
to influence their future course, but to review their substance and ap- 
praise their significance. 

Whatever is lacking to economics through its limited access to the 
exact methods of the natural sciences has to be made good by the ex- 
perience, breadth of knowledge, acumen, and judgment of the individual 
student. A systematic attempt to assess the progress of the various seg- 
ments of economic research and thinking should afford valuable supple- 


ments to the intellectual resources of the individual economist in any 
field. For this reason, the editor of the present survey hopes that its 
character may be such as to lead to the periodic appearance of systematic 
reviews of economics, of even greater merit, in the future. Indeed, the 
subject matter of economics has become so vast and its techniques so 
specialized and difficult, that the day pi] the exhaustive treatise by a 
single Jovian figure, such as Mill, Marshall, or Pigou, may have passed. 
Its place may be taken by compendia of the present sort: less personal, 
less literary, and less unified, perhaps; but— it is to be hoped— less in- 
tuitive, less prescinded, and no less inspiring. 

Howard S. Ellis 


Preface v 

Authors . xi 

Critics xv 


Author: Bernard F. Haley 
Critics: E. H. Chamberlin 
J. M. Clark 

Author: William Fellner 
Critics: Gottfried Haberler 
Alvin H. Hansen 



Author: J. K. Galbraith 
Critics: R. A. Gordon 

A. D. H. Kaplan 


Author: Joe S. Bain 
Critics: Joel P. Dean 

Donald H. Wallace 


Author: Arthur Smithies 
Critics: James K. Hall 

Lawrence H. Seltzer 


Author. Lloyd A. Metzler 
Critics: James W. Angell 
Jacob Viner 

Chapter 7. ECONOMICS OF LABOR 255 

Author: Lloyd G. Reynolds 
Critics: Clark Kerr 

Sumner H. Slichter 




DATA 288 

Author: Carl S. Shoup 
Critics: M. A. Copeland 
E. E. Hagen 

Chapter 9. MONETARY THEORY 314 

Author: Henry H. Villard 
Critics: Edward S. Shaw 
Elmer Wood 


Author: Paul A. Samuelson 
Critics: Friedrich A. Lutz 
Fritz Machlup 

Chapter 11. ECONOMETRICS 388 

Author: Wassily Leontief 
Critics: Joseph A. Schumpeter 
W. Allen Wallis 


Author: Abram Bergson 
Critics: Frank D. Graham 
A. P. Lerner 


Author: David McC. Wright 
Critics: Frank H. Knight 
Paul M. Sweezy 

Index of Names 473 

Index 479 


JOE S. BAIN, Associate Professor of Economics, University of California, 
Berkeley. A.B. University of California at Los Angeles, 1935; Ph.D. Harvard 
University, 1940. Learned societies, etc.: Secretary, Pacific Coast Committee 
on Price Policies, Social Science Research Council, 1942-48. Chief publica- 
tions: Economics of the Pacific Coast Petroleum Industry, three volumes, Uni- 
versity of California Press, Berkeley, 1944-47; Pricing, Distribution and Employ- 
ment, Henry Holt, New York, 1948; War and Post-war Developments in the 
Southern California Petroleum Industry, Haynes Foundation, Los Angeles, 1945; 
'The Profit Rate as a Measure of Monopoly Power," Quarterly Journal of Eco- 
nomics, February 1941, LV, pp. 271—293. 

ABRAM BERGSON, Associate Professor of Economics, Columbia University. 
A.B. Johns Hopkins University, 1933; Ph.D. Harvard University, 1940. Gov- 
ernmental positions: Office of Strategic Services: Economist, 1942—44; Chief, 
Economic Subdivision, Union of Soviet Socialist Republics Division, 1944-46. 
Department of State: Consultant, 1944—46; Member, American Delegation to 
Reparations Conference, Moscow, 1945. Learned societies, etc.: Conference 
on Research in Income and Wealth, National Bureau of Economic Research. 
Chief publications: Structure of Soviet Wages, Harvard University Press, 
Cambridge, Mass., 1944; "A Problem in Soviet Statistics," Review of Economic 
Statistics, November 1947, XXIX, pp. 234—242; "A Reformulation of Welfare 
Economics," Quarterly Journal of Economics, February 1938, LII, pp. 310-334. 

WILLIAM FELLNER, Professor of Economics, University of California, Berkeley. 
Ph.D. Berlin University, 1929. Governmental positions: Consulting Expert, 
Tax Division, Treasury Department, 1945. Chief publications: A Treatise on 
War Inflation, University of California Press, Berkeley, 1942; Monetary Policies 
and Pull Employment, University of California Press, Berkeley, 1946; joint editor, 
Readings in the Theory of Income Distribution, The Blakiston Company, Phila- 
delphia, 1 946; "The Technological Argument of the Stagnation Thesis," Quarterly 
Journal of Economics, August 1941, LV, pp. 636-651; "Prices and Wages under 
Bilateral Monopoly," Quarterly Journal of Economics, August 1947, LXI, pp. 
5°3 - 53 2 ; "Monetary Policies and Hoarding in Periods of Stagnation," Journal of 
Political Economy, June 1943, LI, pp. 192-205. 

J. K. GALBRAITH, Member of the Board of Editors, Fortune Magazine. B.S. 
University of Toronto, 1931; M.S. University of California, 1933; Ph.D. Uni- 


versity of California, 1934. Governmental positions: Economic Adviser, Na- 
tional Defense Advisory Commission, 1940-41. Office of Price Administration: 
Assistant Administrator in Charge of Price Division, 1941—42; Department Ad- 
ministrator, 1942-43. Director of United States Strategic Bombing Survey, 1945. 
Director of Office of Economic Security Policy, State Department, 1946. Learned 
societies, etc.: Fellow, Social Science Research Council, 1937-38. Chief 
publications: with H. D. Dennison, Modern Competition and Business Policy, 
Oxford University Press, New York, 1938; The Economic Effects of the Federal 
Public Works Expenditures, National Resources Planning. Board, United States 
Government Printing Office, Washington, D. C, 1940. 

BERNARD F. HALEY, Professor of Economics, Stanford University. A.B. Stan- 
ford University, 1922; A.M. Harvard University, 1926; Ph.D. Harvard University, 
r 933« Governmental positions: Office of Price Administration: Assistant Re- 
gional Price Executive, 1942; Director, Textile, Leather and Apparel Price Di- 
vision, Washington, 1942-43. Department of State, 1943-45: Assistant Chief, 
Division of Economic Studies; Chief, Commodities Division; Director, Office of 
Economic Affairs; Director, Office of International Trade Policy. Learned 
societies, etc.: Chairman, Pacific Coast Regional Committee of the Social 
Science Research Council, 1941-42. Board of Editors, American Economic Re- 
view, 1938—40. American Economic Association: Executive Committee, 1948—; 
Chairman, Committee on Republications, 1948—. President, Pacific Coast Eco- 
nomic Association, 1942. Chief publications: joint editor, Readings in the 
Theory of Income Distribution, The Blakiston Company, Philadelphia, 1946; 
"The Federal Budget: Economic Consequences of Deficit Financing," American 
Economic Review, February 1941, XXX, pp. 67-107; "The Relation Between 
Cartel Policy and Commodity Agreement Policy," American Economic Review, 
May 1946, XXXVI, pp. 717-734. 

WASSILY LEONTIEF, Professor of Economics, Harvard University. University 
of Leningrad, 1921-25; Ph.D. University of Berlin, 1928. Governmental 
positions: Economic Adviser to the Chinese Government, 1929. Chief, Russian 
Economics Subdivision, Office of Strategic Services, 1943—45. Consultant, Bureau 
of Labor Statistics, 1942-45. Learned societies, etc.: Research Associate with 
National Bureau of Economic Research, 1931. Guggenheim Fellow, 1940. Chief 
publications: The Structure of the American Economy, 1919—192,9, Harvard 
University Press, Cambridge, Mass., 1941; "Essay on Statistical Analysis of Supply 
and Demand," Weltwirtschaftliches Archiv, 1929, pp. 1^-53*; "The Use of In- 
difference Curves in the Analysis of Foreign Trade," Quarterly journal of Eco- 
nomics, May 1933, XLVII, pp. 493-503; "Exports and Imports, Domestic Output 
and Employment," Quarterly journal of Economics, February 1946, LX, pp. 171- 
193; "Introduction to a Theory of the Internal Structure of Functional Relation- 
ships," Econometrica, October 1947, XV, pp. 361-373. 

LLOYD A. METZLER, Associate Professor of Economics, University of Chicago. 
B.S. University of Kansas, 1935; M.A. Harvard University, 1941; Ph.D. Harvard 
University, 1942. Governmental positions: Economist, Office of Strategic 
Services. 1943-44. Economist, Board of Governors of the Federal Reserve Svstem, 


1944-46. Consultant, House of Representatives' Special Committee on Post-war 
Economic Policy and Planning, 1945. Member, Mission to Germany on German 
Currency Reform, American Military Government, 1946. Chief publications: 
"The Transfer Problem Reconsidered, " Journal of Political Economy, June 1942, 
L, pp. 397-414; "Factors Governing the Length of Inventory Cycles," Review of 
Economic Statistics, February 1947, XXIX, pp. 1 — 1 5; "Exchange Rates and the 
International Monetary Fund/' International Monetary Policies, Board of Gov- 
ernors of the Federal Reserve System, Washington, 1947, pp. 1—45. 

LLOYD G. REYNOLDS, Professor of Economics and Associate Director of the 
Labor and Management Center, Yale University. B.A. University of Alberta, 1931; 
M.A. McGill Llniversity, 1933; Ph.D. Harvard University, 1936. Governmental 
positions: Regional Price Executive, Office of Price Administration, 1941-42. 
Public Panel Member and Public Member of the National Appeals Committee, 
National War Labor Board, 1943-45. Consultant, Bureau of the Budget, 1945- 
47. Chief publications: The Control of Competition in Canada, Harvard Uni- 
versity Press, Cambridge, Mass., 1940; Labor and National Defense, Twentieth 
Century Fund, New York, 1941; with C. C. Killingsworth, Trade Union Publica- 
tions, 1850— 1941, three volumes, Johns Hopkins University Press, Baltimore, 
1944 and 1945. 

PAUL A. SAMUELSON, Professor of Economics, Massachusetts Institute of 
Technology. B.A. University of Chicago, 1935; M.A. Harvard University, 1936; 
Ph.D. Harvard University, 1941. Governmental positions: Consultant, Na- 
tional Resources Planning Board, 1941—43. Consultant, War Production Board, 
1945. Learned societies, etc.: American Economic Association, John Bates 
Clark Award, 1947. Chief publications: foundations of Economic Analysis, 
Harvard University Press, Cambridge, Mass., 1948; Economics: An Introductory 
Analysis, McGraw-Hill, New York, 1948; "The Gains from International Trade," 
Canadian Journal of Economics and Political Science, May 1939, V, pp. 195-205; 
"Interactions Between the Multiplier Analysis and the Principle of Acceleration," 
Review of Economic Statistics, May 1939, XXI, pp. 75-78. 

CARL S. SHOUP, Professor of Economics in the School of Business and the 
Faculty of Political Science, Columbia University. A.B. Stanford University, 1924; 
Ph.D. Columbia University, 1939. Governmental positions: New York State 
Special Tax Commission, 1930-35. Consultant, U.S. Treasury Department, 1938—. 
Consultant, Council of Economic Advisers, 1946—. Learned societies, etc.: 
Director of the Survey of Taxation for the Twentieth Century Fund, 1937, Pacing 
the Tax Problem. Chief publications: The Sales Tax in France, Columbia 
University Press, New York, 1930; with Robert M. Haig, The Sales Tax in the 
American States, Columbia University Press, New York, 1934; with Milton 
Friedman and Ruth Mack, Taxing to Prevent Inflation, Columbia University 
Press, New York, 1943; Principles of National Income Analysis, Houghton 
Mifflin, Boston, 1947. 

ARTHUR SMITHIES, Professor of Economics, Harvard University. B.A. Oxford 
University, 1932. Ph.D. Harvard University, T934. Governmental positions: 


Consultant, Board of Economic Warfare, 1942. Consultant, Office of Price Ad- 
ministration, 1942. Bureau of the Budget: Principal Fiscal Analyst, 1943-45; 
Chief Fiscal Analyst, 1945-46; Chief, Economics Branch, 1946. Learned soci- 
eties, etc.: Tasmanian Rhodes Scholar, 1929. Commonwealth Fund Fellow, 
1932. Chief publications: "Equilibrium in Monopolistic Competition," Quar- 
terly Journal of Economics, November 1940, LV, pp. 95-115; "The American 
Economy in the Thirties," American Economic Review, May 1946, XXXVI, pp. 
1 1-27; "Effective Demand and Employment," Chapter XXXIX in The New 
Economics, Seymour Harris, editor, Alfred A. Knopf, New York, 1947. 

HENRY H. VILLARD, Professor of Economics, Hofstra College. A.B. Yale Uni- 
versity, 1932; A.M. Cambridge University, 1939; Ph.D. Columbia University, 
1 94 1. Governmental positions: Senior Economist, Division of Research and 
Statistics, Treasury Department, 1 94 1 . Chief Economist and Branch Chief, Service 
Trade Branch, Office of Price Administration, 1942—43. Economic Adviser, North 
African Economic Board, Department of State, 1943—44. Economist, Board of 
Governors of the Federal Reserve System, 1945-46. Chief publications: 
Deficit Spending and the National Income, Farrar and Rinehart, New York, 1941; 
"The Federal Reserve System's Monetary Policy in 1931 and 1932," journal of 
Political Economy, December 1937, XLV, pp. 721—739; "Dr. Moulton's Estimates 
of Saving and Investment," American Economic Review, September 1937, XXVII, 
pp. 479-489. 

DAVID McC WRIGHT, Professor of Economics, University of Virginia. LL.B. 
University of Virginia, 1935; M.A. Harvard University, 1939; Ph.D. Harvard 
University, 1940. Governmental positions: Consultant, National Resources 
Planning Board, 1943. Chief publications: The Creation of Purchasing Power, 
Harvard University Press, Cambridge, Mass., 1942; The Economics of Disturbance, 
Macmillan Company, New York, 1947; Democracy and Progress, Macmillan Com- 
pany, New York, 1948; "Internal Inconsistency in D. H. Robertson's 'Saving 
and Hoarding Concepts, 1 " Economic Journal, June-September 1941, LI, pp. 


James W. Angell, Columbia University 
E. H. Chamberlin, Harvard University 
J. M. Clark, Columbia University 
M. A. Copeland, Cornell University 
Joel P. Dean, Columbia University 
R. A. Gordon, University of California 
Frank D. Graham, Princeton University 
Gottfried Haberler, Harvard University 
E. E. Hagen, Bureau of the Budget 
James K. Hall, University of Washington 
Alvin H. Hansen, Harvard University 
A. D. H. Kaplan, Brookings Institution 
Clark Kerr, University of California 
Frank H. Knight, University of Chicago 
A. P. Lerner, Roosevelt College 
Friedrich A. Lutz, Princeton University 
Fritz Machlup, Johns Hopkins University 
Joseph A. Schumpeter, Harvard University 
Lawrence H. Seltzer, Wayne University 
Edward S. Shaw, Stanford University 
Sumner H. Slichter, Harvard University 
Paul M. Sweezy, Wilton Centre, N. H. 
Jacob Viner, Princeton University 
Donald H. Wallace, Princeton University 
W. Allen Wallis, University of Chicago 
Elmer Wood, University of Missouri 




Bernard F. Haley 

An attempt is to be made, in the following pages, to review the more 
important contributions to value and distribution theory that have oc- 
curred in the past decade. In view of the volume of the literature that has 
appeared in the English language alone, with reference to these two 
phases of economic theory, this involves a highly selective process. Prob- 
ably no two economists would agree as to all of the material that should 
be included, nor as to the material that, for lack of space, might most 
justifiably be excluded. The selective process must admittedly be a sub- 
jective one. The attempt will be made, however, to include those contri- 
butions that appear to have carried forward the development of theory in 
some significant respect, or that appear likely to stimulate further impor- 
tant theoretical analysis. 

I. Value Theory 

With respect to the theory of demand, the most significant develop- 
ments have been the utilization of the indifference curve technique, and 
related concepts, for a restatement of particular and general equilibrium 
theory; and the kinked demand curve as applied to the case of oligopoly. 
The history of the indifference curve technique certainly antedates the 
period to be covered by the present review. The credit belongs, however, 
to J. R. Hicks and R. G. D. Allen for having made non-mathematically 
trained economists aware of the possibilities of this approach for the im- 
provement of particular and general equilibrium theory. The most impor- 
tant single work in value theory that has appeared during the decade is 
Mr. Hicks' Value and Capital. 1 It has been followed by other able sys- 
tematic treatments of equilibrium theory. 2 

With respect to analysis of the conditions of supply for the enterprise, 
empirical cost studies have suggested the necessity of revising previouslv 
accepted views as to the probable shape of the supply schedule for the 

1 Oxford, 1939. 

"E.g., O. Lange, Price Flexibility and Employment (Bloomington, 1944); J. L. 
Mosaic, General-Equilibrium Theory in International Trade (Bloomington, 1944). 


firm, and have led to a re-examination of the assumptions underlying 
marginal analysis as applied to the behavior of the firm. 

As to particular equilibrium theory, the most important develop- 
ments have had reference to the cases of oligopoly and bilateral monop- 
oly. In general, monopolistic competition theory has been improved bv 
virtue of the increased attention that has been given to variables other 
than price and quantity, such as product variation, variation in location, 
and other manifestations of non-price competition. There has been more 
recognition of the need for tempering broad generalizations with respect 
to the behavior of a small number of variables by consideration for the 
institutional situation and the technical aspects of particular industries. 


( i ) The indifference curve approach to the theory of demand is based 
on the assumption that the individual, in planning his expenditures, so 
determines his outlay on any one commodity that he will be on the 
margin of indifference as between the last small increment of that com- 
modity and an additional alternative increment of any other commodity 
that might be substituted. 3 Assume that the consumer is confronted with 
only two commodities upon which, in varying quantities, he may spend 
his income. In Figure i, let the quantities of the two commodities be 
measured respectively on the OY and OX axes. Then I x and I 2 are indif- 
ference curves. Any combination of X and Y indicated by a point on I l9 
for example, is equally desirable from the point of view of the consumer. 
Furthermore, since I 2 lies farther from O than l lt it is assumed that any 
combination of X and Y indicated by a point on I 2 is preferred to any 
combination of X and Y indicated by a point on l 1% The ratio of exchange 
between X and Y may be shown by the slope of price lines, such as P x , 
P 2 , ... If the individual's income in terms of Y is OM, and if the price 
of X in terms of Y is shown by the slope of P 2 , the diagram shows that the 
consumer would wish to consume OX 2 of X, at a cost to him of FR of Y, 
since at this rate of consumption of X the price line corresponding to his 
income would be tangent to an indifference curve. His rate of consump- 
tion of X 2 places him on the highest indifference curve attainable for 
him, given the income OM and the price P 2 . Given the whole map of the 
individual's indifference curves and given his income, it is possible to 
deduce his demand schedule for X. 4 

3 J. R. Hicks and R. G. D. Allen, "A Reconsideration of the Theory of Value," 
Economica, February and May 1934, I, pp. 52-76, 196-219; H. Schultz, The Theory 
and Measurement of Demand (Chicago, 1938), Ch. 1; J. R. Hicks, Value and Capital, 
Ch. 1, 2. 

4 Analogous to the indifference curve in the theory of demand is the "isoquant" in 
production theory. If, in Figure 1, the two axes measured respectively quantities of two 


One advantage claimed for this approach is that it is not necessary to 
assume that the consumer is capable of cardinally measuring the amount 
of utility attributable to any given increment of a commodity. All that is as- 
sumed is that the individual does weigh the relative desirability of differ- 
ent combinations of quantities of X and Y. Hence the concept of marginal 
utility is replaced by the concept of the marginal rate of substitution. The 
marginal rate of substitution of X for Y is the quantity of Y that would just 

compensate the consumer for the loss of a marginal unit of X. For any com- 
bination of X and Y, it is measured by the slope of the indifference curve 
at the point which represents that combination. When the individual is in 
equilibrium with respect to his consumption of X and Y the marginal rate 
of substitution of X for Y will be equal to the price of X in terms of Y, and 
thus the price line must be tangent to the indifference curve at the equi- 
librium point. 

factors o£ production, X and Y, the curve Ii would be an isoquant, showing the various 
combinations of X and Y that would produce a given output. I 2 would be an isoquant 
with reference to a higher level of output. F. Y. Edgeworth early described a three- 
dimensional figure to illustrate the principle of varying proportions of the factors: ''Con- 
tributions to the Theory of Railway Rates," Economic Journal, September 191 1, XXI, 
pp. 362-363. Use of isoquants has been made by Frisch, Schneider, Hicks, Carlson, 
Boulding, and others. For references, see K. E. Boulding, "The Theory of the Firm," 
American Economic Review, December 1942, XXXII, p. 800. 


Another advantage claimed for this approach is that it makes it easier 
to distinguish between the effect of a change in income and the effect of 
a change in relative prices upon the demand of an individual for a com- 
modity. If the price of X falls relative to the price of Y, the increase in the 
quantity of X taken by the consumer may be broken down into two parts, 
one part attributable to the fact that the consumer's income is in effect 
increased by the fall in the price of X (the income effect), the other part 
attributable to the fact that the consumer will tend to increase his con- 
sumption of X at the expense of other commodities because of the relative 
cheapness of X (the substitution effect). 

Hicks also utilizes the concept of the marginal rate of substitution for 
the definition of substitute and complementary goods so as to avoid the 
assumption of cardinally measurable utility. X and Y are substitute com 
modifies if a decrease in the price of X has the effect of decreasing the 
marginal rate of substitution of Y for money; while X and Y are comple- 
mentary goods if a decrease in the price of X has the reverse effect upon 
the marginal rate of substitution of Y for money. 5 That is, the tendency 
to consume more of X (because of the decrease in the price of X) will be 
accompanied by a tendency to consume less of Y, if X and Y are substi- 
tutes, or more of Y, if X and Y are complements, the price of Y assumed 
unchanged, and quite apart from any income effect of the decrease in the 
price of X. 

Another by-product of the indifference curve analysis has been a re- 
vived interest in the concept of consumer's surplus, since it now becomes 
possible to define the concept without assuming either the cardinal meas- 
urability of utility by the consumer or the constancy of the marginal 
utility of money. In the course of the considerable discussion that has 
occurred, the advantage which a consumer gains from a given price situ- 
ation, or which he obtains from a decrease in price, has been treated as 
measurable in terms of equivalent gains or losses of money income. 6 Once 
the assumption of a constant marginal utility of money is dropped, how- 
ever, it becomes clear that there are several alternative ways of measuring 
consumer's surplus in terms of money income : 

(a) The counterpart of the Marshallian concept: the difference be- 
tween what the consumer actually pays for a given quantity of a com- 

6 Allowance being made for the income effect of the decrease in the price of X, which 
should not be permitted to affect the marginal rate of substitution of Y for money. 

e J. R. Hicks, Value and Capital, pp. 38-41; H. W. Robinson, "Consumer's Surplus 
and Taxation: ex ante or ex post?" South African Journal of Economics, September 
I939> VII, pp. 270-280; A. Henderson, "Consumer's Surplus and the Compensating 
Variation," Review of Economic Studies, February 1941, VIII, pp. 117-121; J. R. Hicks, 
"The Four Consumer's Surpluses," ibid., Winter 1943, XI, pp. 31-41; idem, "The 
Generalised Theory of Consumer's Surplus," ibid., 1945-46, XIII, pp. 68-74. 


modify, and die maximum amount he could have been made to pay for 
it, on an all or nothing basis, without being made worse off than if he 
had not bought any of it at all. 

(b) The compensating variation: the maximum deduction from the 
consumer's income which would leave him in a position still to consume 
some quantity of the commodity without being any worse off than if the 
commodity were not available to him at all. 

(c) The equivalent variation: the amount of added income that 
would compensate the consumer for the loss of the opportunity to pur 
chase any of the commodity. 

If, in Figure i , the individual's money income is OM and the price of 
X is shown by the slope of P 2 , the Marshallian consumer's surplus is RV, 
the difference between FR and FV; the compensating variation is MA; 
and the equivalent variation is MK. 7 

It remains to be seen how useful the new tools will actually turn out to 
be. 8 It should be noted, however, that these new formulations are meas- 
ures of satisfaction expressed in terms of money income, and conse- 
quently their employment for the appraisal of matters of economic policy 
involves nearly all of the risks that made the old concept of consumer's 
surplus of such questionable usefulness. 

This brings us to the more important objections that have been made 
to the indifference curve approach to demand theory. These have mainlv 
to do with the conception of human behavior presupposed by the analysis, 
and with its utilization of an "ordinal" rather than a "cardinal" concept of 
utility. 9 

Of those who object to the conception of human behavior presupposed, 
some maintain that the traditional concept of "cardinal" utility should not 
have been dropped, while others maintain that the new analysis has not 
gone far enough in breaking away from the traditional concept of a ra- 
tional, calculating economic man carefully weighing "amounts" of utility 
or "satisfaction." According to the former, the individual does think in 

" There are corresponding concepts of the increment of consumer's surplus attributable 
to a decrease in price. In addition to the Marshallian concept, there are the price-com- 
pensating variation, the quantity-compensating variation, the price-equivalent variation, 
and the quantity-equivalent variation. It is not possible, however, to examine these fur- 
ther refinements here. 

8 For some possible applications see: J. R. Hicks, "The Rehabilitation of Consumers' 
Surplus," Review of Economic Studies, February 1941, VIII, pp. 1 08-1 16; idem, "Con- 
sumers' Surplus and Index Numbers," ibid., Summer 1942, IX, pp. 126-137. 

y It has also been pointed out that the approach is not one which lends itself to em- 
pirical derivation of the values of the functions involved, any more than is possible with 
the older approach. Nor may the indifference function approach be used as a basis for 
empirical studies of consumption, income, prices, and their interrelationships. W. A. 
Wallis and M. Friedman, "The Empirical Derivation of Indifference Functions," Studies 
in Mathematical Economics and Econometrics (Chicago, 1942), pp. 175-189. 


quantitative terms of the "subjective service" or satisfaction associated 
with successive increments of consumption of a particular good. 10 Hence 
"total satisfaction" is a cardinal magnitude and the individual does have 
a quantitative conception of diminishing marginal utility with respect to 
individual goods and total income. Since the individual does think in 
terms of cardinal utility and increments of total satisfaction, it is a mis- 
take not to base the theory of demand upon these concepts. 

Hicks would answer that there is no objection to retention of the quan- 
titative conception of utility in the theory of demand if anyone cares to 
retain it. However, the notion of cardinal utility is not necessary to the 
explanation of price determination. "Therefore, on the principle of 
Occam's razor, it is better to do without it." 11 

The important question would appear to be: in which way does the 
individual consumer think his way through the problem of allocating his 
income? Does he think in terms of the relative importance of a small in- 
crement in his rate of consumption of one commodity vs. that of an- 
other? Or does he think in terms of an estimated quantity of satisfaction 
to be expected from a small increment in his rate of consumption of a 
particular commodity? 

The question has also been raised, on the other hand, whether the in- 
difference curve approach has not, after all, retained most of the objec- 
tionable features of the psychology presupposed in the older cardinal 
utility approach. 12 Although the theory of demand has been freed from 
dependence upon the assumption that utility is cardinally measurable by 
the individual and the assumption that the marginal utility of money is 
constant, it still seems to be assumed that the individual is able to meas- 
ure his preferences quite precisely, and does so in a coldly rational way. 
There is no recognition of the way in which the individual's preferences 
may be shaped by advertising and other selling methods, no allowance for 
the effects of habit and custom, and no realization that the individual's 
map of indifference curves may be a very short-run phenomenon, subject 
to frequent and possibly capricious change. 

(2) The high degree of simplification of the circumstances affecting 
demand characteristic of the indifference curve approach becomes evi- 
dent when one comes to consider the conditions of demand in situations 

w F. H. Knight, "Realism and Relevance in the Theory of Demand," Journal of Politi- 
cal Economy, December 1944, LII, pp. 289-318. 

11 J. R. Hicks, Value and Capital, p. 18. 

12 P. A. Samuelson, "A Note on the Pure Theory of Consumer's Behaviour," Eco- 
nomica, February 1938, V, pp. 61-62; R. T. Norris, The Theory of Consumer's De- 
mand (New Haven, 1941), Ch. 3; J. M. Clark, "Realism and Relevance in the Theory 
of Demand," Journal of Political Economy, August 1946, LIV, pp. 347-353. 


of oligopoly and monopolistic competition. Mrs. Norris has shown the 
way in which product differentiation has multiplied the number of "com- 
modities" from which the consumer must choose. 13 From this wide vari- 
ety of goods the consumer typically buys only one at a time of any one 
highly specialized and differentiated type of good. Hence the indiffer- 
ence curve apparatus, with its assumption of quantities of homogeneous 
goods measured on the respective axes, does not appear to be a very useful 
tool. 14 The theory of consumer's choice becomes very complicated indeed. 

Most of the recent discussion of demand under conditions of monopo- 
listic competition and oligopoly has, however, had reference to the de- 
mand curve as viewed by the individual concern. 15 What are the different 
types of situation, with respect to demand, that may occur under condi- 
tions of monopolistic competition and oligopoly; and what may be ex- 
pected to be the price policy of the enterprise in each case? 

The first, and most significant, of the developments has resulted from 
a recognition that the sales curve of the firm under conditions of oligopoly 
is not necessarily continuous but may be kinked or have an obtuse angle 
(as viewed from the axes) at the point representative of the current 
price. 16 The upper part of the sales curve is relatively elastic, reflecting 
the fear on the part of the enterprise that if it raises its price its competi- 
tors will not follow suit; while the lower part of the curve is relatively less 
elastic, on the basis that the enterprise expects any price cut it may make 
to be followed promptly by corresponding price cuts by its competitors. 
Such a situation may be particularly frequent in a period of depression, 

33 R. T. Norris, The Theory of Consumer's Demand, Ch. i, 7-9. 

14 Ibid., p. 45. Cf. J. M. Clark, op. ch., p. 351. 

15 There has been some discussion as to whether the demand curve in question is the 
one "imagined" by the enterprise, or is the "real" or "objective" demand curve that the 
enterprise will in fact encounter when it sets its price. (The notion of the "imagined" 
demand curve is credited to N. Kaldor, "Mrs. Robinson's 'Economics of Imperfect Com- 
petition/ " Economica, August 1934, I, pp. 340-341.) As Triffin points out, the relevant 
sales or demand curve is the one which expresses the expectations of the enterprise. It 
may or may not correspond with the actual demand situation as the latter unfolds. Triffin 
also points out that economists writing in this field have usually assumed that the expec- 
tations of the enterprise are always in fact realized. Such an assumption may be appro- 
priate to static analysis, but may not safely be made with respect to a dynamic economy. 
R. Triffin, Monopolistic Competition and General Equilibrium Theory (Cambridge, 
Mass., 1940), pp. 62-67. Some of the consequences of mistaken demand expectations, 
both on the assumption of stationary conditions and on the assumption of dynamic con- 
ditions, have been considered by R. H. Coase, "Some Notes on Monopoly Price," Review 
of Economic Studies, October 1937, V, pp. 17-31; and S. Weintraub, "Monopoly 
Equilibrium and Anticipated Demand," Journal of Political Economy, June 1942, L, pp. 


16 R. L. Hall and C. J. Hitch, "Price Theory and Business Behaviour," Oxford 
Economic Papers, May 1939, No. 2, pp. 22-25; P- M. Sweezy, "Demand under Condi- 
tions of Oligopoly," Journal of Political Economy, August 1939, XLVII, pp. 568-573. 
The development has recendy been reviewed by G. J. Stigler, "The Kinky Oligopoly 
Demand Curve and Rigid Prices," ibid., October 1947, LV, pp. 432-437. 


since at such a time the enterprise might well assume that its rivals were 
operating at less than capacity and were consequently in no mood to lose 
further business as a result of price-cutting by others, nor averse to acquir- 
ing additional business as a result of price increases by others. 

The situation is illustrated in Figure 2, in which the sales curve is 
DD'D", and the corresponding marginal revenue curve is MQRM'. The 
prevailing price is XD'. There is a range of discontinuity in the marginal 
revenue curve, the extent of the discontinuity depending on the differ- 
ence in elasticity of the two segments of the sales curve. 


Figure 2 

This kind of kinked demand curve is by no means the only possibility. 
Sweezy has also suggested the possibility of a kinked demand curve with 
an angle, as viewed from the axes, of more than 180 degrees. 17 In Figure 
2, such a curve is ED'E', with the corresponding marginal revenue curve 
NRQN'. The upper stretch of the curve is now relatively inelastic, indi- 
cating the expectation on the part of the enterprise that price increases by 
it are likely to be followed by similar price increases on the part of rival 
firms. This might well be the expectation in a period in which the con- 
cerns are generally operating at close to capacity output. On the other 
hand, the lower section of the curve is relatively elastic, indicating the 

17 P. M. Sweezy, op. cit., pp. 570-571. Efroymson has explored the possibility further, 
naming this curve the "reflex" curve, in contrast to the "obtusely kinked" curve consid- 
ered earlier. C. W. Efroymson, "A Note on Kinked Demand Curves," American Eco- 
nomic Review, March 1943, XXXIII, pp. 104-107. 


expectation that price decreases by the enterprise are not likely to be fol- 
lowed by rival concerns. This assumption also may be appropriate to a 
period of prosperity; it has also been suggested as appropriate to a situa- 
tion in which price reductions take the form of secret concessions from 
list prices. 18 

Clearly the obtusely kinked and the reflex demand curves are only two 
of several forms that the sales curve of the individual enterprise may take 
under conditions of oligopoly, according to the assumptions that are made 
as to the probable reactions of rivals to the price policy of a given enter- 
prise. 19 The cases considered above, however, appear rather more likely 
to occur than others. 20 

(3) In this review of developments in the theory of demand in recent 
years, some mention should be made of the empirical contributions in 
the field of statistical demand studies. Here the outstanding work of the 
past decade was, of course, the comprehensive and scholarly treatment of 
both the theory of demand and techniques of statistical analysis of de- 
mand provided by the late Henry Schultz. 21 The product of a ten-year 
program of research, this treatise is probably most valuable for its careful 
appraisal of the advantages and disadvantages of the different techniques 
of statistical analysis of demand, at all times considered in the light of 
demand theory. The work also includes, however, actual demand studies 
with respect to about fifteen agricultural commodities. Some of the most 
interesting, although not the most immediately fruitful, of his studies are 
those designed to measure the kind and degree of interrelations of the 
demands for several commodities (such as beef and pork).-- Are the com- 
modities complements or substitutes? What are the cross-elasticities of 
demand? 23 

There were other statistical demand studies during the past decade, 
probably the most important of which were those with respect to automo- 
biles and steel— two of the few cases in which commodities were selected 
with respect to which more complex circumstances must be taken into 

18 As J. M. Clark has suggested in his critique of this chapter, it is hardly likely that a 
firm could, for long, expect to sell its full output secretly at a price lower than that of its 

iU Several other possible types are considered by S. Weintraub, "The Foundations of 
the Demand Curve," American Economic Review, September 1942, XXXII, pp. 547- 

20 The implications of the kinked sales curve for the analysis of particular equilibrium 
under conditions of oligopoly are considered below, pp. 19-20. 

- 1 H. Schultz, op. cit. 

'" Ibid., Ch. 18, 19. 

~ ! What is the effect upon the quantity of beef taken by the market if there occurs a 
slight increase in the price of pork? Or more precisely, what is the relative change in the 
quantity taken of one commodity associated with a very small relative change in the 
price of the other commodity? 


account than in the case of agricultural commodities seasonally pro- 
duced. 24 

Little progress was made, however, in the methodology of statistical 
analysis of demand in the direction of bringing the statistically derived 
demand curve closer to the demand curve used by the economic theorist 
in his analysis. 25 It has to be recognized that the difficulties in the way of 
attaining such an objective may simply be insoluble. The continuing sta- 
tistical studies of demand, however, cannot help but enrich the econo- 
mist's knowledge of the conditions underlying the demand for individual 
commodities studied, and may also lead to improvements in his theoreti- 
cal analysis. 


Developments in the analysis of the conditions of supply for the enter- 
prise have mainly been concerned with the nature of the cost schedule 
for the individual concern. With respect to the short-run situation, these 
developments included: (i) a recognition of the fact that the cost of pro- 
duction of a commodity which constitutes its supply price in the short 
run cannot be regarded as independent of anticipated future prices and 
costs; (2) a considerable range of empirical studies of variations in short- 
run costs for individual enterprises; and (3) partly as a result of these 
empirical cost studies, a recognition of the multiplicity of forms that the 
cost functions of different enterprises may take in different types of tech- 
nological situation. With respect to the long-run situation, (4) the princi- 
pal developments had to do with the optimum size for the enterprise. 

(1 ) The Keynesian concept of user cost is important as a more precise 
statement of the nature of the contribution of economic depreciation to 
prime cost in the short run. 26 The user cost of a unit of output consists of 
two elements: (a) the actual value of the materials used up in producing 
that unit of output (including any materials employed in the operation 
and maintenance of the equipment utilized), flus (b) the reduction in the 
discounted value of the expected future increments of income to be at^ 

21 P. De Wolff, "The Demand for Passenger Cars in the United States," Econometrica, 
April 1938, VI, pp. 1 1 3-1 29; C. F. Roos and V. von Szeliski, "Factors Governing 
Changes in Domestic Automobile Demand," The Dynamics of Automobile Demand 
(New York, 1939), pp. 21-95; H. G. Lewis, "A Statistical Analysis of the Demand for 
Steel, 1919-1938," United States Steel Corporation T.N.E.C. Papers (New York, 
1940), pp. 169-221. 

86 Cf . G. J. Stigler, "The Limitations of Statistical Demand Curves," Journal of the 
American Statistical Association, September 1939, XXXIV, pp. 469-481. 

38 J. M. Keynes, The General Theory of Employment, Interest and Money (London, 
1936), pp. 52-73. Marshall's concept of prime cost included "the extra wear-and-tear of 
plant," but he touched upon it only lightly. A. Marshall, Principles of Economics (Lon- 
don, 1890; 8th ed., 1920), p. 360. 


tributed to the equipment through using it for that unit of output instead 
of leaving it unused (but maintaining it). 27 We are mainly interested 
here in the second of the two elements in user cost, the part of deprecia- 
tion of equipment which is properly to be included in prime cost and 
marginal cost in the short run. 28 

The concept of user cost serves to bring out the dependence of the 
current short-run supply schedule upon expectations as to the future 
course of costs and selling prices. The depreciation element in user cost, 
which enters into supply price, is in the nature of an opportunity cost. 
The cost attributable to the present use of the equipment is the net yield 
that, it is expected, could be obtained instead some time in the future if 
it were not to be used now. 29 Hence marginal cost may change without 
any variation in rate of output or in immediately anticipated costs of ma- 
terials, rates of wages, etc., the change in marginal cost being attributable 
to a change in expectations as to, for example, future wage rates, future 
costs of materials, or future selling price. 

There is some question, however, whether in most situations the de- 
preciation element in user cost is very important. 30 Consequently this re- 
finement in marginal cost analysis should not be overstressed. Further- 
more, the business man's conception of cost of production only very 
roughly approximates at best the conception attributed to him by the 
theoretical economist. 

(2) It was the objective of a group of Oxford economists under the 
leadership of R. L. Hall and C. J. Hitch to discover the way in which 
business men do decide what price to charge and what output to pro- 
duce. 31 A considerable number of entrepreneurs were interviewed, and 
the answers of thirty-eight of these were selected for analysis. The prin- 
cipal conclusion was that most based their selling prices upon some sort 

27 J. M. Keynes, op. cit., p. 53. See also A. P. Lerner, "User Cost and Prime User 
Cost," American Economic Review, March 1943, XXXIII, pp. 1 31-132. 

88 In an integrated economy this depreciation element in user cost would be the only 
element. Consequently the term is sometimes used to refer to the depreciation element 
alone. Cf. P. T. Bauer, ''Notes on Cost," Economica, May 1945, XII, p. 96. J. S. Bain has 
made an excellent analysis of the components of the depreciation function, and of the 
relation between those components which are a part of supply price and those which are 
not. J. S. Bain, "Depression Pricing and the Depreciation Function," Quarterly Journal 
of Economics, August 1937, LI, pp. 705-715. 

29 The dependence of user cost upon expectations as to the future course of costs and 
selling price has been well illustrated from the behavior of the rubber plantations by 
P. T. Bauer, op. cit., pp. 90-100. See also A. C. Neal, "Marginal Cost and Dynamic 
Equilibrium of the Firm," Journal of Political Economy, February 1942, L, pp. 45-64. 

30 Bain shows that, in industries in which the expected rate of obsolescence is high, 
the element of depreciation in user cost may be negligible. J. S. Bain, op. cit., pp. 

31 R. L. Hall and C. J. Hitch, op. cit., pp. 12-45. 


of full average cost including an allowance for profit, and did not think 
in terms of marginal cost or marginal revenue at all. 32 It should be noted 
that none of the cases considered approached even remotely the condi- 
tions of pure competition, and that most of the cases were characterized 
by oligopoly, with or without differentiation of product. It should also be 
noted that "full average cost" was not the same concept in all cases. In 
some cases, it turned out to be simply the price which the price leader 
regarded as covering his full cost; in others, it was a "standard" cost figure 
published by the trade association; in still others, it was actually com- 
puted independently by each concern. For some of the enterprises, how- 
ever, it was a fairly flexible figure; some admitted "that they might charge 
more in periods of exceptionally high demand, and a greater number that 
they might charge less in periods of exceptionally depressed demand." 33 
Furthermore, even when the individual concerns computed their costs 
independently, it appears that they varied their respective profit margins 
included in the "full cost" figure "so that approximately the same prices 
for similar products would rule within the 'group' of competing pro- 
ducers." 31 

As Machlup has shown, there appears to have been considerable atten- 
tion paid by several of the entrepreneurs to what the economist has in 
mind when he uses the terms elasticity of demand and marginal reve- 
nue. 35 Machlup, however, is distrustful of the whole procedure of this 
study, since he does not believe that much information of value can be 
obtained from business men's answers to questions about their reasons for 
charging the prices they charge. The stress upon "full cost" may simply 
be a rationalization or justification of the price actually charged, or it may 
refer to the price arrived at by agreement of the members of the industry, 
or the price which the individual entrepreneur believes will yield him his 
fair share of the business without inducing his competitors to expand 
their businesses at his expense and without attracting new firms to the 
industry. 30 

32 Cf. Committee on Price Determination, Conference on Price Research, Cost Be- 
havior and Price Policy (New York, 1943), p. 286. The Committee discovered that the 
full cost principle appeared to be in fairly wide use among firms in the United States. 

33 R. L. Hall and C. J. Hitch, op. tit,, p. 19. 

35 F. Machlup, "Marginal Analysis and Empirical Research," American Economic Re- 
view, September 1946, XXXVI, pp. 536-547. See also replies to Machlup by R. A. 
Lester, "Marginalism, Minimum Wages, and Labor Markets," ibid., March 1947, 
XXXVII, pp. 135-142; and H. M. Oliver, Jr., "Marginal Theory and Business Be- 
havior," ibid., June, 1947, XXXVII, pp. 375-383. 

30 A study by R. A. Lester, "Shortcomings of Marginal Analysis for Wage-Employ- 
ment Problems," American Economic Review, March 1946, XXXVI, pp. 63-82, based 
on questionnaires producing responses from thirty-six manufacturers, was interpreted to 
show that most of the replying firms believed that they were subject to decreasing varia- 


Inadequate though these studies were in many respects, they may have 
served the useful purpose of making economic theorists more skeptical of 
the applicability, without great caution, of the traditional apparatus of 
cost curves, particularly in the cases of oligopoly and monopolistic com- 
petition. Those responsible for the studies have relied so heavily upon the 
answers of their respondents alone, however, that it probably w^ould be 
unwise to give too much weight to their conclusions until these studies 
have been supplemented by further research in the behavior and motiva- 
tion of entrepreneurs with respect to price policy. 

Another important source of knowledge of the costs of the enterprise 
and, like the work of Lester and of Hall and Hitch, a stimulus to the re- 
examination of theoretical concepts, has been the series of statistical 
studies of costs which have appeared since 1933. Statistically derived cost 
curves designed to show the relation between variations in output and 
cost per unit of product have been constructed from accounting data for 
a considerable variety of industries. 37 Statistical procedures have been 
employed for the elimination of the effects upon cost of influences other 
than variations in output, such as changes in the prices of the factors, 
technological changes, and the lag of output behind costs; and the resid- 
ual relationship between total cost and output has then been measured 
by a line of regression of best fit. In most cases the resulting cost curve has 
been representative, of necessity, of the behavior of costs for the aggregate 
of products of the concern rather than for a single undifferentiated com- 

With only very few exceptions, the line of regression of best fit has 
been a straight line, and consequently the conclusion has been drawn 
that marginal cost within the range of variations of output covered by 
these studies has probably been constant. Such a conclusion is of course 
widely at variance with the economist's U-shaped marginal cost curve for 

ble costs per unit of output within the range of 70 to 100 per cent of capacity. As 
Machlup has pointed out, the usefulness of the study must be seriously questioned, how- 
ever, since Lester's questionnaire did not include any definition of "capacity," and it is 
therefore not possible, unfortunately, to determine the range of outputs the respondents 
might have had in mind when they reported decreasing variable costs. Other aspects of 
the replies indicate that little confidence can be placed in the evidence they represent. 
F. Machlup, op. cit., pp. 550-552. See also R. A. Lester, "Marginalism, Minimum 
Wages, and Labor Markets," ibid., March 1947, XXXVII, pp. 138-139. 

" 7 For example: J. P. Dean, Statistical Determination of Cost with Special Reference 
to Marginal Cost (Chicago, 1936); T. O. Yntema, "An Analysis of Steel Prices, Volume 
and Costs: Controlling Limitations on Price Reductions," United States Steel Corpora- 
tion T.N.E.C. Papers (New York, 1940), I, pp. 223-323. For appraisal of these and 
other contributions, see: H. Staehle, "The Measurement of Statistical Cost Functions: 
An Appraisal of Some Recent Contributions," American Economic Review, June 1942, 
XXXII, pp. 321-333; Committee on Price Determination, Conference on Price Research, 
Of. cit., Ch. 5. 


the individual concern in the short run as inferred from the principle of 
varying proportions. 

Although there have now been several of these statistical cost studies 
that have yielded a constant marginal cost curve within the range of ob- 
servations included, economists have not found the evidence by any 
means conclusive. In the first place, it has been pointed out, the account- 
ing data and some of the methods of analysis employed (such as the com- 
putation of depreciation on a straight-line basis, the allocation of over- 
head over time, and the use of annual average figures for output) have 
introduced an important bias toward linearity. 38 Furthermore, the selec- 
tion of the straight line as the line of regression of best fit for the cor- 
rected figures for cost and output may not in every case be justified. Very 
slight deviations from linearity in this statistically determined total cost 
function would yield a marginal cost curve of the traditional variety, and 
such a slightly curvilinear function might fit the corrected figures about 
as well as a straight line. 39 

It consequently appears premature to conclude that marginal cost, at 
least in the case of manufacturing enterprises, is typically constant in- 
stead of an increasing function of output. Furthermore, since these 
studies have not in general covered a range of outputs that included those 
in the neighborhood of capacity and beyond, they have been responsible 
for no doubts whatever as to the inevitability of rising marginal cost if the 
concern expands its output beyond "capacity." On the other hand, ques- 
tionable though the conclusions of the studies may be, they have led to 
a re-examination of the assumptions on the basis of which the traditional 
marginal cost curve for the enterprise has been drawn, and to the con- 
sideration of possible situations in which marginal cost might, at least for 
outputs not departing veiy far from normal, be nearly constant. 

(3) The conventional U-shaped marginal cost curve is probably ap- 
plicable to the case in which the fixed plant is indivisible but is adapt- 
able to the utilization of successive small increments of the variable 
factors. This, however, may be a special case. More likely cases will be 
those characterized by some degree of divisibility of fixed plant and 
some flexibility. 40 The greater the number of identical machines in the 

38 H. Staehle, op. cit., pp. 328-330; C. A. Smith, "The Cost-Output Relation for the 
U.S. Steel Corporation," Review of Economic Statistics, November 1942, XXIV, pp. 
168-171; Committee on Price Determination, Conference on Price Research, op. cit., 
pp. 96-102. 

39 R. Ruggles, "The Concept of Linear Total Cost-Output Regressions," American 
Economic Review, June 1941, XXXI, pp. 332-335. 

40 G. Stigler, "Production and Distribution in the Short Run," journal of Political 
Economy, June 1939, XLVII, pp. 305-322; also in W. Fellner and B. F. Haley, eds\, 


plant, the greater the divisibility. The more divisible is the fixed plant, 
and the less each unit is adaptable to variations in the quantity of the 
variable factors, the greater will be the tendency toward constant mar- 
ginal cost in the range of outputs short of full utilization, since increases 
in output up to that point will be obtained simply by bringing into opera- 
tion successive additional units of the fixed plant together with the appro- 
priate additional quantities of labor and materials to employ the new 
units. Of course perfect divisibility of fixed plant is probably a limiting 
case, but the existence of some degree of divisibility, Stigler suggests, 
should affect the shape of the marginal cost curve. Similarly, the presence 
of flexibility in the plant may work in the same direction. Flexibility of 
operations is obtained by various techniques that reduce the extent to 
which average cost increases for outputs greater than or less than the op- 
timum. 41 One method of increasing flexibility is to increase the divisibil- 
ity of fixed plant beyond the point that would be desirable if the plant 
were expected to operate continuously at the optimum. Another method 
is to reduce fixed plant relative to variable factors. Flexibility thus will 
probably involve a cost : the level of the average cost curve may be some- 
what higher than for a less flexible plant designed to operate at the lowest 
possible minimum cost output. Variations in average cost with variations 
in output should not be so great, however, as in the less flexible plant, and 
consequently the marginal cost curve should also be flatter. 

(4) Analysis of the determinants of the optimum size of the firm has 
profited from several contributions. First, there is the principle of increas- 
ing risk, which has been advanced as one source of limitation upon in- 
crease in scale of enterprise. 42 The larger the entrepreneur's own invest- 
ment in the firm, the more is his total financial situation endangered in 
the event that the firm should fail. Furthermore, the larger his invest- 
ment, the greater is the danger he incurs from sacrifice of liquidity. The 
applicability of this limitation upon size appears restricted, however, to 
the case of the individual proprietorship, and to a less extent, the partner- 
ship form of enterprise. The corporation, with its feature of limited lia- 
bility of stockholders and with its ability to sell additional issues of stock 
to many investors, does not appear to be confronted with this sort of ob- 
stacle to increase in size. 43 

Readings in the Theory of Income Distribution (Philadelphia and Toronto, 1946), pp. 

41 Ibid., pp. 314-317. Cf. a subsequent discussion by J. Dean, Statistical Cost Func- 
tions of a Hosiery Mill (Chicago, 1941), pp. 7-15. 

42 M. Kalecki, ''The Principle of Increasing Risk," Economica, November 1937, IV, 
pp. 440-447. 

43 N. S. Buchanan and R. D. Calkins, "A Comment on Mr. Kalecki's 'Principle of 
Increasing Risk,'" Economica, November 1938, V, pp. 455-458. Kalecki, in his reply 


A limitation upon size somewhat similar to that suggested by Kaleeki is 
the less than perfect elasticity of supply of enterprise that may character- 
ize the attitude of entrepreneurs. 44 It has been generally assumed that the 
entrepreneur regularly seeks to maximize profits, and that he expands his 
enterprise to the point beyond which no increment of profit is to be ob- 
tained by further increase in output. The entrepreneur may not be indif- 
ferent, however, to other aspects of increase in size, such as the prestige 
and power that come from conduct of a large enterprise, or on the other 
hand the stress and strain, the heavy responsibility, and other possible dis- 
advantages of bigness. 

In the case of the large corporation it is particularly likely that the 
assumption that the entrepreneur seeks to maximize profit may not be 
valid.* 8 One important case in point is the corporation whose entrepre- 
neur regards retention of control as at least as important as the maxi- 
mizing of profits. In this case, growth of the firm toward the technically 
optimum size may be checked by a fear of partial or complete loss of con- 
trol as a consequence of resort to outsiders for additional capital. 40 


1 he principal contributions to particular equilibrium theory in recent 
years have had reference to the case of oligopoly and, to a less extent, to 
the case of bilateral monopoly. 17 Little has been added to previous analy- 
sis of the cases of pure competition and of the large group case under 
monopolistic competition. In this section, consideration will be given to: 

Qihid., pp. 459-460), maintains that there is a limit to the extent to which a corporation 
can raise new capital through common stock issues, and that this limit may affect the 
size of the corporation in the same way that increasing risk may affect the size of the 
individual proprietorship or partnership. 

44 T. de Scitovszky, "A Note on Profit Maximisation and Its Implications," Review of 
Economic Studies, Winter 1943, XL pp. 57-60; K. E. Boulding, "The Incidence of a 
Profits Tax," American Economic Review, September 1944, XXXIV, pp. 567-572. Cf. 
B. Higgins, "Elements of Indeterminacy in the Theory of Non-Perfect Competition," 
ibid., September 1939, XXIX, pp. 476-479. 

45 M. W. Reder, "A Reconsideration of the Marginal Productivity Theory," journal of 
Political Economy, October 1947, LV, pp. 450-458. Cf. E. G. Nourse, Price Making in 
a Democracy (Washington, 1944), pp. 98-»o5- 

46 M. W. Reder, op. cit., pp. 455-457. J. M. Clark, "Toward a Concept of Workable 
Competition," American Economic Review, June 1940, XXX, pp. 248-249, has found 
evidence to indicate that other factors than size appear to be principally responsible for 
differences in cost between plants, and that, consequently, there is typically no definite 
"optimum size" of plant but rather "a wide optimum range of size." 

47 Unfortunately, space does not permit more than mention of the considerable im- 
provement in the diagrammatic apparatus for the exposition of the equilibrium conditions 
with respect to selling costs for the firm. The principal contributions are those of K. E. 
Boulding, Economic Analysis (New York and London, 1941), pp. 578-588, 616-618; 
and N. S. Buchanan, "Advertising Expenditures: A Suggested Treatment," Journal of 
Political Economy, August 1942, L, pp. 537-557. 


CO the position that particular equilibrium analysis is inappropriate to 
cases of oligopoly and monopolistic competition; (2) recent developments 
in the theory of oligopoly; and (3) recent developments in the theory of 
bilateral monopoly. 

( 1 ) Serious doubts have been raised as to whether it is appropriate or 
fruitful to attempt to develop particular equilibrium analysis for cases of 
oligopoly and monopolistic competition, particularly for a differentiated 
commodity. 48 Under pure competition an industry produces a homoge- 
neous commodity, each producer of which realizes that his product is a 
perfect substitute for the product of each of his competitors in the in- 
dustry. Under monopolistic competition, however, if the product is differ- 
entiated, the concept of an industry becomes necessarily vague. Each 
competitor is producing a somewhat unique product. He competes in 
varying degree with all other firms in the economy, and the competitive- 
ness of technologically similar products has no peculiar significance that 
justifies theoretical analysis of particular equilibrium for the "industry." 
"In the general pure theory of value, the group and the industry are use- 
less concepts." 49 It is urged, therefore, that attention be directed immedi- 
ately to the problem of general equilibrium. Interdependence in selling 
should be measured simply by the cross-elasticities of demand, without 
regard to the technological similarities to be discerned among products. 

Since, however, different brands or makes of the same "commodity" in 
general do compete with one another somewhat more closely than do dif- 
ferent commodities, there appears to be empirical justification for retain- 
ing the concept of the industry, very much as Chamberlin has, as a highly 
flexible tool, and for continuing to employ a form of particular equilib- 
rium analysis in monopolistic competition theory. Retention of the con- 
cept of the industry need in no way impede progress in the development 
of a general equilibrium approach such as Triffin urged, but which, thus 
far at least, has not proved particularly fruitful. 

(2) "The theory of oligopoly has been aptly described as a ticket of 
admission to institutional economics." 5 " One important development in 
the literature of oligopoly theory in recent years has been the increased 
emphasis given to the institutional aspects of the problem. 01 The stage of 

4S R. Triffin, op. cit., pp. 78-108. 

49 Ibid., p. 89. 

50 E. S. Mason, "Price and Production Policies of Large-Scale Enterprise," American 
Economic Review, March 1939, Proceedings, XXIX, pp. 64-65. 

nl For example, M. Abramovitz, "Monopolistic Selling in a Changing Economy," 
Quarterly journal of Economics, February 1938, LII, pp. 191-214; E. S. Mason, op. cit., 
pp. 61-74; J- M. Clark, "Toward a Concept of Workable Competition," loc. cit., pp. 
241-256; W. H. JNicholls, A Theoretical Analysis of Imperfect Competition with Special 
Application to the Agricultural Industries (Ames, 1941), Ch. 4-11. 


development of the industry will to some extent determine entrepreneurs' 
knowledge of market conditions, the probable reactions to be expected 
from rival enterprises, and in general the intensity of competition. The 
different sizes of firms in the industry, the motivation of those in control 
(whether they are ambitious to expand their share of the market, or are 
content to let sleeping dogs lie), and the existence or lack of price leader- 
ship are other circumstances that have to be taken into account. Facilities 
for the spread of knowledge of the market may be a stabilizing element, 
reducing discrimination or price-cutting. In general, the tendency has 
been to recognize that each oligopolistic industry is to some extent 
unique, and not too much is to be gained from the multiplication of 
theoretical models. 

It has also been suggested that the traditional approach to oligopoly 
makes a mistake in assuming that the firm necessarily seeks only to max- 
imize profit. Other equally important objectives may necessitate a compro- 
mise with the profit objective. For example, the entrepreneur may be 
equally motivated by a desire for "security." 52 Furthermore, oligopoly 
theory must take into account the struggle for position that is constantly 
taking place, or threatening to take place. "To write a short manual on 
the Principles of Oligopolistic War would be a very important attempt 
towards a new approach to this aspect of price theory; . . ." 53 

As to the theoretical models that have been developed on the assump- 
tion that the firm does seek to maximize profits, the principal basis for 
classification has been the reaction which A expects from B as a result of 
what A does, not only as to price, but also as to the technological quality 
of his product, his outlay for selling efforts, and other circumstances per- 
taining to his product. Whatever the reaction of B assumed by A, further 
classification may be based on whether A's forecast of B's reaction is as- 
sumed to be correct or not. The classic Cournot and Bertrand models 
were based respectively on two different assumptions as to B's reactions to 
A's price policy, but in neither case was B's reaction in fact the one ex- 
pected by A. These models, and others like them, involve of necessity a 
form of sequence analysis of which there obviously could be a large num- 
ber of variations. During the period under review, not very much has 
been added to the analysis of this type of case. 54 

52 K. W. Rothschild, 'Price Theory and Oligopoly," Economic Journal, September 
1947, LVII, pp. 299-320. 

53 Ibid., p. 307. Reference should also be made, although no more than a reference is 
possible in spite of its importance, to the new approach to the analysis of oligopoly sug- 
gested by J. von Neumann and O. Morgenstern, Theory of Games and Economic Be- 
havior (Princeton, 1944). 

54 Mention should be made, however, of A. Smithies, "Equilibrium in Monopolistic 
Competition," Quarterly Journal of Economics, November 1940, LV, pp. 95-115. 


Much more attention has been directed to the type of model in which 
it is assumed that each oligopolist is, with somewhat more sophistication, 
capable of correctly anticipating the reactions of his rivals to his price 
policy. One such case, developed previously to the kinked demand curve 
cases, is that of the industry in which each producer correctly assumes 
that his rivals would match any price cut. If the cost functions of the 
rivals are assumed to be equal, the outcome is the monopoly price. 55 

By far the most attention, however, has been given to the case in which 
each enterprise conceives its sales curve to be obtusely kinked, with the 
consequence that its marginal revenue curve is discontinuous. In Figure 
1, the case is illustrated by the sales curve DD'D", and the correspond- 
ing marginal revenue curve MQRM'. 56 On the assumption that the en- 
terprise seeks to maximize its profit, it is in equilibrium at the current 
price XD' as long as the marginal cost curve cuts the marginal revenue 
curve at any point within the range of discontinuity. Hence the price 
XD' is likely to be very stable. 57 

An increase in demand for the product is likely to result in the upper 
part of the demand curve becoming less elastic and the lower part more 
elastic, since the firms will now be operating more nearly at capacity. 58 
Hence the range of discontinuity will be reduced. The reverse will occur 
in a period of declining demand. 

The kink may under certain circumstances temporarily disappear. If, 
for instance, there should occur a general increase in wage rates or mate- 
rials cost, affecting the whole industry, it might appear at first sight as 
though the consequent upward shift in the short-run supply schedule 
might not alter selling price and output, provided that the supply sched- 
ule continued to cut the marginal revenue curve within the zone of dis- 
continuity. If, however, the increase in costs is general for the whole in- 
dustry, the enterprise will now no longer regard the upper part of its sales 

55 M. Abramovitz, op. cit., pp. 193-195; G. J. Stigler, "Notes on the Theory of 
Duopoly," Journal of Political Economy, August 1940, XLVIII, pp. 528-530. Other 
assumptions are involved, including probably the implicit assumption that each producer 
correctly anticipates that his rivals would also meet any price increase. The case in 
which the cost functions are not equal has also been considered by T. Kristensen, "A 
Note on Duopoly," Review of Economic Studies, October 1938, VI, pp. 56-59, and by 
Stigler, loc. cit. 

50 Above, p. 8. 

51 It has been suggested that the concept of the obtusely kinked sales curve is useful in 
the explanation of price rigidity, and in the analysis of the effects of open price agree- 
ments, cutthroat competition, and industrial racketeering. M. Bronfenbrenner, "Applica- 
tions of the Discontinuous Oligopoly Demand Curve," Journal of Political Economy, 
June 1940, XLVIII, pp. 420-427. See also G. J. Stigler, "The Kinky Oligopoly Demand 
Curve and Rigid Prices," loc. cit., pp. 432-449. 

58 P. M. Sweezy, op. cit., p. 571. 


curve to te relatively elastic, since it will now expect other concerns to 
follow a price increase. 59 For the period of readjustment, the concern 
thinks of its sales curve as continuous and unkinked; but immediately the 
upward adjustment of price for the industry has been completed, the 
concern will revert to thinking in terms of a kinked sales curve, the kink 
now occurring of course at the new price. 

The analysis does not appear reversible. If, in a period of depression 
for example, a decrease in wages or materials cost should occur, the enter- 
prise is likely to resist downward revision of selling price since it will 
continue to think in terms of an inelastic sales curve with respect to 
lower prices on the assumption that other enterprises will promptly fol- 
low price cuts. 60 Price discrimination, particularly in the form of secret 
price concessions, may gradually bring about price reduction on a more 
general scale. Or the reduction of the prices of closely competitive com- 
modities may finally necessitate a general reduction of prices in the 

A somewhat different type of model is concerned with the case in 
which each producer assumes that his rivals will insist on maintaining 
their respective "shares of the market." 61 If it is assumed that the cost 
functions are equal, the outcome must be the monopoly price. If marginal 
costs differ, the lower-cost firms may set slightly lower prices than the 
higher-cost firms. If the different firms' conceptions of their shares of the 
market add up to more than the total output that can be disposed of at 
the prices set, the situation is rendered unstable. 

In the case of differentiated products, uniformity of price is not so 
likely to be the outcome as in the case of homogeneous commodities. In 
addition to price and output, variations in quality of product, selling 
costs, and other manifestations of non-price competition are involved in 
the process of reaching equilibrium for the industry. The most interest- 
ing contributions have been those which have taken differences in loca- 
tion of the enterprises constituting an industry as representative of 
differentiation of product. 

The starting point for most of this sort of analysis is Hotelling's early 
article on the subject. 02 His analysis was devoted to a hypothetical com- 
munity of consumers equally distributed along a straight line and each 

59 L. G. Reynolds, "Relations between Wage Rates, Costs, and Prices," American 
Economic Review, March 1942, Proceedings, XXXII, pp. 276-281; also in Readings 
in the Theory of Income Distribution, pp. 296-302. 

60 Ibid., p. 280. 

01 G. J. Stigler, "Notes on the Theory of Duopoly," loc. cit., pp. 530-531. 

02 H. Hotelling, "Stability in Competition," Economic Journal, March 1929, XXXIX, 
pp. 41-57- 


having a perfectly inelastic demand per time unit for one unit of the 
commodity produced by two enterprises, A and B. The cost of trans- 
portation per unit of distance is constant, the cost of production per unit 
is also constant (at zero), and transport costs are paid by the buyer. The 
location of A being determined before B comes on the scene, B assumes 
that A will not change his location, and that A's price policy will be 
unaffected by the price which B charges. 

Hotelling's case was remote from reality, but it has provided a point 
of departure for a number of writers who have successively brought the 
assumptions more nearly into accordance with the situation actually con- 
fronting a small number of concerns competing spatially in the sale of 
a commodity. 63 In particular, the introduction of a negatively inclined 
demand schedule for the commodity at each point in the linear market, 
recognition of the ability of each competitor to move his location at will, 
and consideration of various more likely price policies on the part of the 
rival concerns have made possible considerable improvements in the 
analysis. Further variations in the fundamental assumptions have pro- 
duced a considerable number of models that collectively contribute sub- 
stantially to an understanding of the complexities of oligopoly and mo- 
nopolistic competition under conditions of differentiation of product. It 
is clear that the existence of transportation costs alone is sufficient to 
render competition less than pure. Furthermore the level of trans- 
portation costs has much to do with determining the spatial distribution 
of firms and the level of mill prices relative to cost of production. Even 
though the prices are quoted f.o.b. mill, there may be considerable 
freight absorption by the rival producers. If, however, each, producer 
absorbs the whole of the freight to his customers within a given area 
(if, that is, prices are quoted on a free delivery basis), the pattern of 
distribution of the firms will be quite different. The analysis is not 
essentially affected by varying the assumption with regard to cost of 
production, whether marginal costs are assumed to rise or to fall with 
increases in output; but the assumption of different cost functions for 
the different concerns will result in different prices f.o.b. mill. The in- 
troduction of quality differentiation (in addition to spatial differenti- 
ation) and discrimination tends to increase scale of plant and to reduce 

63 Some of the more recent contributions: A. P. Lerner and H. W. Singer, "Some 
Notes on Duopoly and Spatial Competition," Journal of Political Economy, April 1937 
XLV, pp. 145-186; M. A. Copeland, "Competing Products and Monopolistic Compe 
tition," Quarterly Journal of Economics, November 1940, LV, pp. 1-35; A. Smithies 
"Optimum Location in Spatial Competition," Journal of Political Economy, June 1941 
XLIX, pp. 423-439; W. A. Lewis, "Competition in Retail Trade," Economica, No 
vembcr 1945, XII, pp. 202-234. 



monopoly profits. 64 In addition to throwing considerable light on the 
theory of location of industry and the effects of spatial differentiation 
upon price and output, the analysis has provided an approach to the ex- 
amination of basing-point systems. 65 

(3) The theory of bilateral monopoly has reference to the situation 
in which a monopolistic seller confronts a monopsonistic buyer. The 
theory has focused attention mainly on three supposedly distinct types 
of case, which may be briefly summarized with reference to Figure 3. 

Figure 3 

In the first case, the seller (A) is relatively weak, with the result that 
the buyer (B) fixes the price (OYJ so that it is equal to A's marginal cost 
or supply price (shown by MC A ) for that amount sold (OX a ) for which 
the increment in supply price (shown by MC/) is equal to B's marginal 
net product for this same amount of raw material purchased. (At the 
output OX a , MC/ intersects MNP B . The supply price for this output, as 
shown by MC A , is GY^) 

64 Copeland has paid particular attention to this case. He has also stressed the point 
that brand and quality competition are not precisely analogous to spatial competition. 
Of. cit., pp. 17-28. 

65 See, for example, A. Smithies, "Aspects of the Basing-Point System," American 
"Economic Review, December 1942, XXXII, pp. 705-726; V. A. Mund, "Monopolistic 
Competition Theory and Public Price Policy," ibid., pp. 727-743; J. M. Clark, "Imper- 
fect Competition Theory and Basing-Point Problems," ibid., June 1943, XXXIII, pp. 


In the second case, the buyer is relatively weak, with the result that 
the seller (A) fixes the price (OY 2 ) so that it is equal to B's marginal net 
product for that amount of raw material purchased (OX b ) whose incre- 
ment in marginal net product (shown by MNP P /) is equal to the seller's 
supply price (shown by MC A ). 

In the third case, A and B are conceived to maximize their combined 
profits, the amount of raw material which changes hands being that 
amount (OX e ) for which A's marginal cost is equal to the marginal net 
product to B (i.e., the amount of X shown by the intersection of MC A 
and MNP B ). The price paid would be indeterminate between OY 3 , 
equal to the average net product of OX c for B, and OY 4 , equal to A's 
average cost for the amount sold. 60 

The first and second cases have generally been regarded as limiting 
cases, between which most actual situations of different degrees of rela- 
tive bargaining power may be expected to fall. Leontief and Fellner, 
however, have drawn attention to the fact that either B, in the first case, 
or A, in the second case, could push his advantage still further by deter- 
mining the quantity to be exchanged, on an all or nothing basis, as well 
as the price. In the first case, the result would be that B would fix the 
amount to be purchased at OX e (on an all or nothing basis), and the 
price at OY 4 , thus reducing A's profit to zero. In the second case, A 
would fix the amount to be sold at OX c (on an all or nothing basis) and 
the price at OY 3 , thus reducing B's profit to zero. 67 

It now becomes clear, as Fellner has pointed out, that what we really 
have is the third case; the first two cases do not appear to be genuine 
cases at all. Unless there should be some institutional obstacle which 
would prevent an all or nothing contract, both parties would gain by 
moving from the first or the second situation to the third, in which out- 
put would be increased and the joint profit maximized. 68 In the case of 
product markets there does not appear to be any such institutional ob- 
stacle, and consequently it is not likely that either of the first two cases 
will be found in these markets. Bilateral monopoly tends to establish 

08 The three cases have been summarized by G. Tintner, "Note on the Problem of 
Bilateral Monopoly," Journal of Political Economy, April 1939, XLVII, pp. 263-267; 
and W. Fellner, "Prices and Wages under Bilateral Monopoly," Quarterly Journal of 
Economics, August 1947, LXI, pp. 503-509. Both authors provide bibliographies of the 
earlier literature on the subject. In addition to the diagrams to be found in these two 
articles, ingenious diagrams have been developed by A. M. Henderson, "A Further 
Note on the Problem of Bilateral Monopolv," Journal of Political Economy, April 1940, 
XLVIII, pp. 238-243; and A. J. Nichol, "Monopoly Supply and Monopsony Demand," 
ibid., December 1942, L, pp. 861-879. 

87 W. Leontief, "The Pure Theory of the Guaranteed Annual Wage Contract," Journal 
of Political Economy, February 1946, LIV, pp. 77-79; W. Fellner, op. cit., pp. 506, 

6 ^ W. Fellner, op. cit., pp. 524-528. 


output at the level which maximizes the joint profit, the price paid by 
the monopsonistic buyer being indeterminate between two limits, the 
upper of which would eliminate profit for the buyer, the lower of which 
would eliminate profit for the seller. 69 


A revival of interest in analysis of the general equilibrium of the 
economy is largely to be credited to J. R. Hicks. 70 His contribution has 
been of some importance simply because he was willing to develop his 
analysis in a form accessible to the non-mathematical economist. Much 
more important, however, is the fact that he brought together in a single 
systematic treatment the theory of general equilibrium under static con- 
ditions and the analysis of the economy under dynamic conditions. This 
has been called 'one of the most important achievements of modern 
economics." 71 

In fact, Hicks* interest in the analysis of general equilibrium under 
static conditions was primarily as a point of departure for the study of 
the instability which characterizes the dynamic economy. It is not sur- 
prising therefore that his principal contribution to static general equilib- 
rium theory is his examination of the conditions necessary for the stability 
of the system. 

There are several self-imposed limitations upon the analysis: (i) he 
assumes an economy in which competition is pure; (2) he abstracts from 
State interference in economic affairs; and (3) he abstracts from capital 
and interest, saving and investment, and speculation. 72 The third of these 
limitations is dropped in the later analysis of the dynamic economy. 

Four markets are involved: (1) the market for products; (2) the 
market for factors; (3) the market for direct services; and (4) the market 
for intermediate products. In each case, Hicks recognizes that, if sta- 
bility exists at all, it may be either perfect or imperfect stability. A 
market is perfectly stable if a fall in price below equilibrium results 
in an excess of demand over supply at the new price even after all 
other prices in the system have been readjusted to the new price. A 
market is imperfectly stable if a fall in price below equilibrium results 

W9 Applications of the theory of bilateral monopoly to the wage bargain will be con- 
sidered below, pp. 34-36. 

70 Value and Ca-pital, Ch. 4, 5, 8. 

71 L. A. Metzler, "Stability of Multiple Markets: The Hicks Conditions," Eco- 
nometrica, October 1945, XIII, p. 277. 

72 J. R. Hicks, Value and Capital, pp. 99-100. M. W. Reder has discussed the stability 
conditions under circumstances in which competition is not pure: "Monopolistic Cor 
petition and the Stability Conditions," Review of Economic Studies, February 194 
VIII, pp. 122-125. 



in an excess of demand over supply at the new price only after all other 
prices have been readjusted to the new price. 73 In the former case, the 
market for a single commodity is not only stable taken by itself, but 
this stability is not disturbed by repercussions from changes in other 
prices resulting from a change in the price of the first commodity. In the 
latter case, the market for the first commodity is not stable taken by 
itself, but it is rendered stable by the repercussions from changes in other 
prices induced by the change in the price of the first commodity. 

In the case of the market for products, Hicks' analysis led him to the 
conclusion that, if the market for a commodity was stable taken by itself, 
it was unlikely to be rendered unstable by repercussions from other price 
changes. Furthermore, the only element likely to render the market for 
a commodity unstable, taken by itself, was the income effect of the fall 
in price, and this was unlikely to be sufficiently important to affect 
stability in the case of products markets. 74 With regard to the case of 
imperfect stability, he was inclined to doubt whether the market for 
a commodity, unstable by itself, would be rendered stable by repercus- 
sions from other markets, although conceivably a large income effect in 
related markets might have this effect. 75 On the whole, he was inclined 
to conclude that a multiple product market was likely to be perfectly 

Of the other three types of market, the market for direct services be- 
haves similarly to the market for products. Markets for intermediate prod- 
ucts involve no income effects, and are therefore even more likely to be 
perfectly stable than markets for products and direct services. In the case 
of factor markets, however, such as the labor market, there is likely to be a 
considerable income effect in addition to the substitution effect of a fall in 
price, and this income effect may be a source of instability. Other sta- 
bilizing elements in the system, however, are probably more than ade- 
quate to offset such unstabilizing income effects as may exist. 76 

Hicks reasoned that if each individual market could be presumed to 
be stable, the whole complex of markets constituting the economic 
system would also be stable. This inference has been seriously criticized 

73 J. R. Hicks, Value and Capital, p. 67. 

7 * A second source of instability, "extreme complementarity," was stressed in Value 
and Capital, but Hicks has since concluded that this part of his analysis was erroneous. 
See his "Consumers' Surplus and Index Numbers," he. cit., p. 133, note; and his "Re- 
cent Contributions to General Equilibrium Economics," Economica, November 1945, 
XII, p. 236. 

75 Idem, Value and Capital, p. 72. 

TO Ibid., p. 103. A good summary of Hicks' analysis of the conditions of stability is 
to be found in F. Machlup, "Professor Hicks' Statics," Quarterly Journal of Economics, 
February 1940, LIV, pp. 284-293. See also O. Lange, Price Flexibility and Employment, 
pp. 91-109; and J. L. Mosak, General-Equilibrium Theory in International Trade, Ch. 2. 


by Samuelson, and later by Metzler. 77 Metzler has summarized the 
criticism as follows: 

It cannot be assumed, as in the Hicks analysis, that when the price of one com- 
modity is out of equilibrium the prices of all other commodities are either un- 
changed or are instantaneously adjusted to their new equilibria. For this reason, 
the Hicks stability conditions cannot be accepted unless it is shown that they are 
related to the stability of a true dynamic system. The errors of the Hicks method 
were first demonstrated by Samuelson in his pioneer article on the significance of 
dynamics to static analysis. It was there shown that imperfect stability, in the Hicks 
sense, is neither a necessary nor a sufficient condition for true dynamic stability. 
An example was given of a dynamic system which was unstable despite the fact 
that it was imperfectly stable in the Hicks sense. Another example was given of 
a system which was dynamically stable even though it was neither perfectly nor 
imperfectly stable according to Professor Hicks's definitions. In a later note, Profes- 
sor Samuelson demonstrated that even perfect stability is insufficient to insure true 
dynamic stability under all circumstances. 78 

The criticism appears sound, and leads directly to the conclusion 
that the analysis of the stability conditions of equilibrium can only prop- 
erly proceed as a part of the development of a dynamic economics. 79 

II. Distribution Theory 

Distribution theory continued to be limited in the main to analysis of 
the determinants of the fer unit rate of remuneration of the factors of 
production. Although such analysis should contribute to an understand- 
ing of the determinants of the corresponding functional shares of the 
national income, little progress is to be reported with respect to this higher 
stage of distribution analysis. The impact of the work of Keynes and his 
followers, however, may in time have the effect of directing more atten- 
tion to this latter problem. 

The concept of marginal productivity has continued to play a promi- 
nent part in the structure of distribution theory. It has, however, been 
subjected to vigorous attack with respect to its usefulness in the expla- 
nation of the demand for labor. Furthermore, proponents of the liquidity 
preference doctrine of interest do not regard the marginal productivity 
of capital as a determinant of the interest rate, although they do attribute 

77 P. A. Samuelson, "The Stability of Equilibrium: Comparative Statics and Dynam- 
ics," Econometrica, April 1941, IX, pp. m-112; idem, "The Relation Between Hicksian 
Stability and True Dynamic Stability," ibid,, July-October 1944, XII, pp. 256-257: 
L. A. Metzler, of. cit., p. 279. 

78 L. A. Metzler, loc. cit. 

79 Thus Lange considers the effect upon stability of the economy of the relative speeds 
of adjustment in the different markets. O. Lange, op. cit., pp. 94~99- Cf. also L. A. 
Metzler, op. cit., pp. 279-285. 


to it the role of determining the amount of investment, given the rate 
of interest. 

It is with respect to distribution theory rather than value theory that 
the impact of Keynesian doctrine has been important. The liquidity 
preference theory of interest is a case in point. Another closely related 
case is Keynes* treatment of saving. Economists already had had some 
doubts about the reliability of the concept of a positively inclined supply 
schedule of savings, treated as identical with the supply schedule of 
capital. Keynes strengthened those doubts, and gained general accept- 
ance for the view that the relation between the rate of savings and the 
rate of flow of income is a more reliable relation, and probably a more 
useful one as well. The general theory of employment has also brought 
home to the economist the fact that the demand for labor, in the complex 
market in which the structure of wages is determined, depends in part 
upon the level of income which depends in part upon the level of employ- 
ment, which is one of the variables supposed to be determined by the 
demand and supply of labor. So the determination of wages cannot be 
isolated from the numerous variables responsible for the determination 
of the level of employment and income. Finally, Keynes' work provided 
an important stimulus to more careful examination of the behavior of 
real and money wage rates in the short run or the business cycle. 

The theory of rent has received relatively little attention. The tend- 
ency has been to associate both rent and profit with the working of the 
dynamic economy. A few contributions of genuine merit during the past 
decade may be credited with producing some progress toward clarifi- 
cation of the concept of profit, of the entrepreneurial function, and of 
the relation between the two. 

Reference was made, at the beginning of this section, to the fact that 
little attention has been given, as yet, to the problem of the determinants 
of the proportion of the national product going to each of the functional 
shares. There are two contributions, however, which should receive 
mention. Kalecki undertook, on a rather fragile statistical basis, to in- 
vestigate the relative share of manual labor in the national income. 80 
He arrived at the conclusion that the relative share of manual labor 
tends to be affected adversely by ( i ) an increase in the (Lerner) degree 
of monopoly, 81 and (2) an increase in raw material prices relative to the 

80 M. Kalecki, "The Distribution of the National Income," Essays in the Theory of 
Economic Fluctuations (London, 1939), pp. 13-41. Cf. an earlier version of the same 
article, "The Determinants of Distribution of the National Income," Econometrica, 
April 1938, VI, pp. 97-112. 

81 The degree of monopoly, according to Lerner, is the ratio of price minus marginal 
cost to price. If marginal cost is equal to marginal revenue, the degree of monopoly is 


wages bill, and vice versa. 82 Although it is not difficult to believe that 
these are important determinants of the proportion of wage income to 
total national income, the analysis is certainly far from satisfactory, as 
Kalecki would probably be the first to agree. 83 

With particular regard to the proportion of wages to income paid out 
in the short run (over the cycle), Dunlop has found that the ratio varied 
considerably from year to year in the period 1919-37. His analysis of the 
reasons for these variations stressed: "(a) extent of fluctuation in out- 
put, (b) the shape of the short-run labor cost function, (c) the relative 
price movements of variable factors and the possibilities of short-run 
substitution, (d) impact of the absolute fluctuation in variable factor 
prices on product prices, (e) the magnitude of technical change, and 
CO the elasticity of product demand for the enterprise." 84 


The Demand for Labor. CO The utilization of the marginal pro- 
ductivity theory for the development of a systematic theory of exploita- 
tion under conditions of monopoly and monopsony has been followed by 
some further contributions in this same area. Defining exploitation as "the 
payment to labor of a wage less than its marginal revenue product," Bloom 
has turned up several types of exploitation in addition to those earlier 
developed. 85 First, there is the situation in which it is costly to change 
price, perhaps because of the additional advertising expense that would 
be involved in informing prospective buyers of the change. Under these 
circumstances a fall in wages may not be followed by any change in 
price unless the new rate of wages is below the marginal revenue product 

the reciprocal of the elasticity of demand. A. P. Lerner, "The Concept of Monopoly and 
the Measurement of Monopoly Power," Review of Economic Studies, June 1934, I, 
pp. 157-175. 

82 See also M. Kalecki, "A Theory of Long-Run Distribution of the Product of Indus- 
try," Oxford Economic Papers, June 1941, No. 5, pp. 31—41, in which he considers the 
determinants of the proportion of wages to value added by an industry. The principal 
ones operative in the long run are: utilization of equipment, ratio of average wage to 
material cost, "quantitative and qualitative divergencies in the investment activity of 
various firms," changes in technique, degree of monopoly, degree of oligopoly, and the 
rate of prime selling cost. 

83 In addition to other criticisms of Kalecki's analysis, J. T. Dunlop points out that 
the part played by the degree of monopoly cannot properly be considered "causal," since 
the degree of monopoly is defined as the gap between price and marginal labor costs 
expressed as a ratio to price, and the proportion of the value product going to labor 
is the ratio of labor cost to price. J. T. Dunlop, Wage Determination under Trade 
Unions (New York, 1944), p. 187, note. 

84 Ibid., pp. 176, 187. 

85 G. F. Bloom, "A Reconsideration of the Theory of Exploitation," Quarterly Journal 
of Economics, May 1941, LV, pp. 413-442; also in Readings in the Theory of Income 
Distribution, pp. 245-277. 


by more than the cost of changing price. The case would appear limited 
to the mature oligopoly not confronted by a strong union, and Bloom 
therefore concludes that this type of exploitation will tend to disappear. 

Second, there is the case in which labor receives less than its marginal 
revenue product because the firm is confronted by a discontinuous de- 
mand and marginal revenue curve for its product. If the marginal cost 
curve cuts the marginal revenue curve within the latter's discontinuous 
range (QR in Figure 2, p. 8) this means that the rate of wages is less than 
the marginal revenue product, and exploitation exists. 86 

(2) Hicks' classification of inventions, it will be recalled, was based 
upon the effect of the invention upon the ratio of the marginal pro- 
ductivities of the factors (labor and capital), the amounts of the factors 
assumed unchanged. 87 A labor-saving invention was one the effect of 
which was to increase the marginal product of capital relative to that of 
labor; a capital-saving invention had the reverse effect, and a neutral 
invention had no effect upon the ratio between the marginal pro- 
ductivities of labor and capital. 

Mrs. Robinson has made the point that the ultimate effect of an in- 
vention upon the relative shares of labor and capital depends not only 
on the immediate effect of the invention upon the relative marginal 
productivities of the factors but also upon the elasticity of substitution 
which prevails as, for example, the supply of capital increases relative 
to the supply of labor in order to restore equilibrium in the capital 
market. 88 Thus the immediate effect of a labor-saving invention may be 
to increase the marginal productivity of capital relative to that of 
labor, and thus to increase the relative share of capital. If the rate 
of interest is assumed constant, the rise in the marginal productivity 
of capital, however, will presumably result in an increase in investment. 
The increase in the amount of capital relative to the amount of labor 
will now further change the relative marginal productivities of the two 
factors. If the elasticity of substitution is less than unity, for example, 
the increase in the ratio of capital to labor will tend to reduce the ratio 
of the marginal productivity of capital to that of labor and thus tend to 
offset the initial adverse effect of the labor-saving invention upon the 
share of labor. 

66 This case was earlier discussed by R. F. Mikesell, "Oligopoly and the Short-Run 
Demand for Labor," Quarterly Journal of Economics, November 1940, LV, pp. 1 61-166. 
It is also interesting to note in passing that the concept of the kinked demand curve 
inspired M. Bronfenbrenner to develop a correspondingly kinked supply curve for labor 
under conditions of oligopsony and a highly conventional wage structure. See his 
"Applications of the Discontinuous Oligopoly Demand Curve," he. cit., pp. 426-427. 

87 J. R. Hicks, The Theory of Wages (London, 1932), pp. 1 21-130. 

88 J. Robinson, Essays in the Theory of Employment (London, 1937), pp. 132-136. 


Mrs. Robinson is thus led to suggest a classification of inventions 
slightly different from Hicks'. 89 If the ultimate effect of an invention 
is to leave the relative shares unchanged, the rate of interest being as- 
sumed constant, the invention is a neutral one; if the ultimate effect is 
to increase the share of capital, the invention is a capital-saving one; in 
the reverse case, it is a labor-saving one. 90 

G. F. Bloom has reported upon a series of field surveys on the nature 
of invention, and has offered some comments with regard to Hicks' analy- 
sis of the invention process. 91 Hicks had, in addition to his classification 
of inventions considered by Mrs. Robinson, a further classification into 
"induced" and "autonomous" inventions, the former being the result of 
a change in the relative prices of the factors. 92 Bloom's survey of business 
experience with inventions convinced him: (i) that relatively few 
inventions fit Hicks' definition of an induced invention; (2) that most 
labor-saving inventions are to be explained, not by changes in relative fac- 
tor prices, as Hicks had maintained, but by the persistently high price of 
labor, and that this is the principal reason for the predominance of labor- 
saving inventions; (3) that, contrary to Hicks, the very labor-saving 
invention (which would have been profitable even without any change 
in relative factor prices, and which may result in a reduction of the 
absolute share of labor) is probably quite common; and (4) that, if 
sufficient time is permitted to elapse for the full effect of an invention to 
be obtained, the ultimate result will almost always be an increase in the 
real wage of labor, even though the initial effect is a reduction in the 
absolute share of labor. 

O. Lange has considered the total effect of an innovation (he employs 
this term rather than invention) on the marginal cost of output and the 
marginal physical productivity of the input planned for the whole cur- 
rent or future period which the firm takes into consideration. 93 Thus he 
has output-neutral, output-increasing, or output-decreasing innovations on 

89 Idem, "The Classification of Inventions," Review of Economic Studies, February 
*938, V, pp. 139-142; also in Readings in the Theory of Income Distribution, pp. 175- 

90 Alternatively, she defines a neutral invention as one which leaves the ratio of capital 
to product (the average productivity of capital) unchanged, after capital has increased 
in response to the initial increase in the marginal productivity of capital. A capital- 
saving invention is one which reduces the average productivity of capital; and a labor- 
saving invention is the opposite. The same definitions are also formulated in terms of 
the effect upon the elasticity of the average productivity curve for capital. 

91 G. F. Bloom, "A Note on Hicks's Theory of Invention," American Economic Re- 
view, March 1946, XXXVI, pp. 83-96. 

92 J. R. Hicks, The Theory of Wages, p. 125. . . 

93 O. Lange, "A Note on Innovations," Review of Economic Statistics, February 1943, 
XXV, pp. 19-25; also in Readings in the Theory of Income Distribution, pp. 181-196. 
Cf. also idem, Price Flexibility and Employment, Ch. 12. 


the basis of the effect upon marginal cost, and factor-neutral, factor-using, 
and factor-saving innovations on the basis of the effect of the innovation 
upon the demand for input factors. An innovation may at the same time 
be output-increasing and factor-saving, etc., with regard to all factors at all 
dates. A special case which he considers, however, is the oligopoly case in 
which the demand curve is kinked. Only innovations which reduce 
marginal cost greatly would have the effect of increasing output, since 
the marginal cost curve would have to be lowered sufficiently to fall be- 
low the range of discontinuity in the marginal revenue curve. 94 Hence 
oligopoly exerts a selective effect in favor of factor-saving innovations, and 
against output-increasing innovations, except when the latter may greatly 
reduce marginal cost. 

(3) For very much the same reasons that doubts have been raised 
about the assumption that management seeks to equate marginal cost 
and marginal revenue, the part played by marginal productivity in the 
analysis of the demand for labor has been subjected to severe criticism. 95 
Lester, in the survey of business men's opinions mentioned earlier, 96 
obtained answers from fifty-six firms as to the circumstances they regarded 
as most important in determining the number of individuals they re- 
spectively employed. Of the various circumstances suggested (market 
demand, wage rates, non-labor costs, profits, production techniques), 
market demand was selected as the only circumstance by half the firms, 
and it was given heavy weight by all of the others. Very little weight 
was given to "the level of wage rates or changes in the level of wage 
rates." Lester therefore concludes that most employers do not think of 
their demand for labor as a function of the wage rate, as the marginal 
productivity approach to the demand for labor would appear to imply, 
but rather as a function of expected sales. 97 One reason, he suggests, that 
business men do not respond to an increase in wage rates by curtailing 
employment is that they associate a reduction in output with an increase 
in variable cost per unit, as mentioned earlier. 98 Another reason is that, 

94 Cf. Figure 2, above, p. 8. Lange also considers the case of monopsony in which 
the supply curve of the factor is kinked. Cf. above, p. 29. In this case, factor-neutral 
innovations would be favored, because of the range of discontinuity in the marginal 
expenditure curve. 

95 R. A. Lester, "Shortcomings of Marginal Analysis for Wage-Employment Problems," 
loc. cit., pp. 63-82; idem, "Marginalism, Minimum Wages, and Labor Markets," loc. cit., 
pp. 135-148. 

98 Above, pp. 12-13. 

97 Machlup points to defects in Lester's questionnaire. He also quite rightly observes 
that the items, "market demand," "non-lalx>r costs," "production techniques," are all 
elements affecting marginal productivity. F. Machlup, "Marginal Analysis and Empiri- 
cal Research," loc. cit., pp. 548-550. 

98 Above, pp. 12-13. 


under existing techniques of production, the extent to which capital 
can be substituted for labor in a given enterprise is ordinarily very limited. 
He finds support for his conclusions in a study of the answers of forty- 
three southern firms to a question as to their probable reaction to an in- 
crease in their wage rates relative to those paid by competitors in other 
areas. Of the various possible reactions from which respondents might 
choose, the most frequently mentioned were improvements in efficiency 
through better management, incentives, etc., and the introduction of 
labor-saving machinery. "Reduce production by deliberately curtailing 
output" was mentioned by only four." 

(4) It is interesting to note that critics of the marginal productivity 
approach as well as its defenders made no reference to the considerable 
number of statistical studies of production as a function of the relative 
quantities of capital and labor, fathered mainly by Douglas and authors 
associated with him, that have continued to appear over the last decade. 100 
One possible reason is the fact that these studies have been rather se- 
verely criticized with respect to the quality of data available, the statistical 
methods employed, and particularly the interpretation of the results. 

The Labor Market. The most important institutional influence on 
the supply of labor, and thus on wage rates, comes from the organization 
of labor. It has been argued by some that existing wage theory is almost 
completely inadequate in the face of the complexities of the present-day 
labor market. 101 Contributions to the analysis of the part played by union 
organizations in the determination of wage rates have been made, how- 
ever, with respect to the following types of labor market situation: (1) 
the open shop industry in which there is no employer discrimination; (2) 

S9 In addition to differing with Lester as to the extent to which the proportion of 
factors can be altered, even in an enterprise already built, Machlup points out that, when 
competition is not pure, the firm's reaction to an increase in wage rates frequently will 
be, not a "deliberate" decrease in output, but an increase in price to a level at which, 
as it turns out, the quantity that can be sold is reduced. Hence the low score given to 
deliberate curtailment is not of any significance. On the other hand, the high scores 
given to the introduction of labor-saving machinery and to price-product changes are 
quite consistent with the marginal productivity approach. F. Machlup, op. cit., pp. 552- 
553. See also Lester's reply, "Marginalism, Minimum Wages, and Labor Markets," 
loc. cit., pp. 135-142. See also G. J. Stigler, "Professor Lester and the Marginalists," 
American Economic Review, March 1947, XXXVII, pp. 154-157. 

100 M. L. Handsaker and P. H. Douglas, "The Theory of Marginal Productivity Tested 
by Data for Manufacturing in Victoria," Quarterly Journal of Economics, November 
1937 and February 1938, LII, pp. 1-36, 215-254; P. H. Douglas and G. Gunn, "Further 
Measurements of Marginal Productivity," ibid., May 1940, LIV, pp. 399-428; and 
numerous others. 

101 For example, R. A. Lester, "Reflections on the 'Labor Monopoly' Issue," Journal 
of Political Economy, December 1947, LV, p. 513; A. M. Ross, "The Dynamics of 
Wage Determination under Collective Bargaining," American Economic Review, Decem- 
ber 1947, XXXVII, pp. 793-798. 


the closed shop industry; (3) the situation in which there is employer 
discrimination against union labor; (4) the case of bilateral monopoly. 102 

(1) In the case of the open shop industry in which there is no em- 
ployer discrimination against union labor, let it be assumed that the 
union seeks to set that wage rate that would maximize the income of its 
employed members (i.e., the area under the demand curve for union 
labor). 10:! If it is further assumed that, since there is no discrimination, 
the demand for union labor at all wage rates is a fixed fraction of the 
total demand for labor, then the union's policy would maximize the in- 
come of all labor employed. If the rate of wages fixed by the union is 
above the equilibrium level that would prevail in the absence of union 
action, there will be unemployment which will be shared by union and 
non-union workers alike. 

(2) If the union enforces a closed or preferential shop, its wage policy 
will be the same as before, but in addition it will seek to provide the 
total of those employed. In the case of the preferential shop, unemploy- 
ment will be concentrated in the non-union group. In the case of the 
closed shop, the union may restrict membership to a number less than 
adequate to provide the employment which would maximize the income 
of labor, thus changing its policy to one of maximizing income per 

(3) There are two kinds of employer discrimination against union 
labor: (a) the employer may give a preference to non-union workers at 
any given wage; (b) the employer may decrease the rate of wages he is 
willing to offer for any given quantity of labor when the labor is union- 
ized, as compared with what he would be willing to offer for that quantity 
of labor if it were not unionized. 104 Both kinds may exist simultaneously. 

If the first kind of discrimination exists, the union cannot do better 

102 The first three are considered by M. Bronfenbrenner, "The Economics of Collective 
Bargaining," Quarterly Journal of Economics, August 1939, LIII, pp. 535-561; the 
fourth, by W. Fellner, "Prices and Wages under Bilateral Monopoly," ibid., August 
1947, LXI, pp. 503-532; cf. the same author's Monetary Policies and Full Employment 
(Berkeley, 1946), pp. 103-m. 

1,13 Dunlop has pointed out that unions may have many important objectives of wage 
policy other than the maximizing of income: e.g., the promotion of membership, the 
allocation of available work, control of the rate of introduction of technical innovations. 
J. T. Dunlop, "Wage Policies of Trade Unions," American Economic Review, March 
1942, Proceeding, XXXII, pp. 290-294; also in Readings in the Theory of Income 
Distribution, pp. 336-341. See also idem, Wage Determination under Trade Unions, 
PP- n 45~54- 

. 1(H It is assumed that both kinds of labor are paid the same rate of wages, but since 
the employer thinks of the union worker as costing him more than the non-union worker 
(because union workers are "troublesome" in some sense), the marginal productivity of 
the union workers employed must, as an offset, be correspondinglv higher than the uni- 
form rate of wages. 


than seek to maximize the area under the demand curve for union labor, 
which will now equal the demand curve for all labor minus, at each wage 
rate, the supply of non-union labor available at that wage rate. If the 
supply of union labor is a relatively high proportion of the total supply, 
the union may still be able to increase the income of its employed mem- 
bers by raising the wage rate above the equilibrium level. The full 
burden of any resulting unemployment, however, would be borne by the 
union membership. 

The effect of the second kind of discrimination, if it is assumed that 
the supply of non-union labor is not adequate to satisfy the demand for 
labor at all wage rates, is to reduce the effective demand for labor below 
the level that would otherwise prevail, and thus to reduce both the quan- 
tity of employment and the rate of wages. The union may still raise the 
rate of wages above the equilibrium level, but the optimum employment 
(from the union's point of view) will be less than it would be in the 
absence of discrimination, to the detriment of both union and non-union 
workers. 105 

(4) When there is a strong union confronted by a single large em- 
ployer or employers' association, the theory of bilateral monopoly is ap- 
plicable. 106 The application of the theory to the labor market, however, 
differs from its application to a commodity market since in the case of a 
unionized supply of labor, the concept of a supply schedule is inappro- 
priate. 107 It may be assumed, however, that there is some level of wages 
below which the union would not accept employment for its members; 
and it also may be assumed that the union, to an extent which varies, 
weighs against one another the advantage of higher wages for its mem- 
bers and the disadvantage of increased unemployment that may accom- 
pany higher wages. 

Analogous to application of the theory to commodity markets, three 
cases may be considered: (a) The employer is sufficiently strong to fix 
the rate of wages. In this case the rate of wages may be fixed at the lowest 
level at which members of the union would be willing to accept em- 

(b) The union is sufficiently strong to fix the rate of wages. If it has 
no regard for the effect upon employment, it will fix the wage at OYx, in 
Figure 4, equal to the maximum average net product of labor (ANP). A 
vague fear of unemployment may temper union policy, however, and 

105 Bronfenbrenner also considers the case in which union labor is paid a different 
wage from that paid non-union labor. Op. cit., pp. 551-561. 

106 Cf . above, pp. 22-24. The analysis which follows is Fellner's. 

107 There are special circumstances, however, which Fellner considers, in which the 
supply schedule of labor is significant. 


result in a wage rate somewhat lower. If it is union policy to give equal 
weight to wage rate and employment, the situation may be pictured by 
an indifference map, the indifference curves l l9 I 2 , I 3 , showing that wages 
and employment are to some degree substitutes for one another. Under 
these circumstances, the union will, if it is sufficiently strong, fix the 
wage rate at OY 2 , the rate at which the marginal net product curve 
(MNP) is tangent to an indifference curve (I 2 ). 

Figure 4 

(c) As in the case of product markets, either of the two parties, if suf- 
ficiently strong, may push its advantage to the full by insisting not only 
on the most advantageous rate of wages but also on the most advanta- 
geous amount of employment, on an all or nothing basis. A case in point 
is the guaranteed annual wage contract. 108 If the union were the stronger 
party, it might insist on the wage rate OY 3 (indicated by the tangency of 
ANP to an indifference curve, I 3 ), at the same time requiring a guarantee 

ws Cf. W. Leontief, "The Pure Theory of the Guaranteed Annual Wage Contract," 
loc. cit., pp. 76-70 


of the corresponding amount of employment, OX 3 . If, however, the em- 
ployer were the stronger, he might drive the wage rate to the lowest possi- 
ble indifference curve (any combination of wage rate and employment ly- 
ing below this indifference curve being unacceptable to labor), which thus 
becomes, in a sense, the supply curve of labor as well as the average cost 
curve of labor to the employer. Let this lowest possible indifference curve 
be Ij, and let MC be drawn marginally to it, showing the marginal cost of 
labor to the employer. Then the employer will seek to equate the marginal 
cost of labor to his marginal net product. If he pushes his advantage to the 
full, he will require an amount of employment OX 4 , on an all or nothing 
basis, at the wage rate OY 4 . At this wage, the marginal cost of labor to the 
employer will be equal to his marginal net product for the amount of em- 
ployment OX4, and labor will be receiving its supply price for this same 
amount of employment. 

It is clear that, as in the case of the all or nothing contract in a com- 
modity market, this type of contract has advantages both for the union 
and the employer, provided that the union's indifference map is concave. 
From the point of view of the employer, however, there is a substantially 
increased risk involved in the fixed employment type of contract. Since 
the wage contract is likely to cover a considerable period, the disadvan- 
tage of the uncertainty involved may outweigh the bargaining advantage 
of the all or nothing type of contract. Similarly, from the point of view 
of the employees, the element of uncertainty for the employer in effect 
reduces the level of his marginal net product curve. Consequently there 
may be situations in which the union, even though it possesses the domi- 
nant position in the bargaining, would not find it to its advantage to insist 
on a contract that guaranteed a minimum volume of employment. 

The Short Period. Mr. Keynes' treatment of the subject of wages in 
his General Theory was probably directly responsible for the exceedingly 
active discussion of short-run wage theory in the years that have fol- 
lowed. 109 There have really been two closely related branches to this dis- 
cussion, one branch concerned with Mr. Keynes' assertion that, in gen- 
eral, as output increases or decreases, with given organization, equipment 
and technique, money wages will ordinarily also increase or decrease, but 
real wages will move in the opposite direction to money wages; the other 

109 J. M. Keynes, The General Theory of Employment, Interest and Money, pp. 5- 
17, 257-271. From the extended literature that has developed, the following are only 
a few of the contributions: A. C. Pigou, "Real and Money Wage Rates in Relation to 
Unemployment," Economic journal, September 1937, XL VII, pp. 405-422; L. Tarshis, 
"Real Wages in the United States and Great Britain," Canadian journal of Economics 
and Political Science, August 1938, IV, pp. 362-376; J. T. Dunlop, "The Movement of 
Real and Money Wage Rates," Economic journal, September 1938, XLVIII, pp. 413- 
434; W. Fellner, Monetary Policies and "Full Employment, pp. 94-103, 1 09-1 11. 


branch concerned with the view that, subject to certain qualifications, 
"there may exist no expedient by which labour as a whole can reduce its 
real wage to a given figure by making revised money bargains with the 
entrepreneurs." 110 

The first of these two propositions clearly was based on the assumption 
that, with given organization, equipment, and technique, the supply 
curves of individual enterprises would be positively inclined in the short 
run, and the further assumption of pure competition. Under these cir- 
cumstances, the increase in money wages that might ordinarily be ex- 
pected to accompany an expansion in output would be more than 
matched by the increase in product prices, with the consequence that real 
wages would decline. The reverse sequence would apply in the case of a 
decrease in output. 111 The fact that real and money wages have not, in 
fact, behaved this way has suggested the necessity of modifying the un- 
derlying assumptions, and of taking into consideration circumstances not 
included in the Keynesian analysis. Out of this discussion there have 
emerged the following propositions with respect to the variables affecting 
real wages as output and money wages increase: (i) Money wages 
usually begin to rise while there is still considerable unemployment, and 
long before normal capacity output has been reached. On the other hand, 
marginal cost may be nearly constant for a considerable range of outputs 
and until peak production is approached. (2) There may be a decrease in 
the degree of monopsony as the business situation improves, as a result, 
for example, of increased strength of the unions. (3) As money wages 
rise there will tend to be, up to a point at least, an increase in the average 
productivity of labor. (4) The prices of factors other than labor that enter 
into variable cost may rise relatively slowly. (5) The degree of monopoly 
may decrease as business improves. All of these circumstances, if present, 
would affect real wages favorably as output increased, and might fre- 
quently more than offset the considerations that Keynes had in mind. 11 " 1 

Mr. Keynes' second proposition is the more important. It has reference 
to the effects of a policy of lowering money wages in a period of depres- 
sion as a means of increasing employment. 113 On the assumption of pure 

110 J. M. Keynes, op. cit., p. 13. 

111 Attention was directed to the proposition, first, by the showing that, historically, 
real and money wages have not in fact behaved in accordance with Mr. Keynes' expecta- 
tion. L. Tarshis, op. cit.; J. T. Dunlop, "The Movement of Real and Money Wage 
Rates," loc. cit.; L. Tarshis, "Changes in Real and Money Wages," Economic Journal, 
March 1939, XLIX, pp. 150-154; also in Readings in the Theory of Income Distribu- 
tion, pp. 330-335. 

"* The same analysis is applicable, inversely, to the case of declining output and 
decreasing money wages. 

113 The subject had, of course, been considered at length previously in the literature 


competition and highly flexible prices, he held that a decrease in money 
wages, in a period of unemployment, will ordinarily result only in a cor- 
responding reduction of prices with no increase in output or employ- 
ment. 114 As in the former case, most of the later contributions to the 
discussion have had to do with considerations not involved in Mr. Keynes' 
presentation of the matter. The principal propositions that have been ad- 
vanced are the following: (1) Prices will not necessarily fall in propor- 
tion to the decrease in wage rates. In the first place, even under conditions 
of competition, if the supply schedules of commodities are positively in- 
clined at all, price would in any event fall less than in proportion even 
though wages were the whole of prime costs. If, in addition, prime costs 
include a considerable proportion of costs other than wages, the decline 
of prices will be still less in proportion to the decline in wages. In the 
second place, if competition is less than pure, prices may be "sticky" and 
may fall much less than the decrease in marginal costs— the extent of the 
fall depending essentially on the degree of monopoly, which is likely to 
be increasing. The net effect on employment is uncertain, but it might 
turn out to be a decrease. 

(2) On the other hand, as money wages decrease there may, even in 
the short run, be some substitution of labor for other factors. (3) The 
reduction in money wages may favorably affect investment expenditure 
which is not related to consumer demand (which would not be expected 
to increase) but to expected investment activity in future periods. (4) 
Any such favorable effect upon employment may, however, be dampened 
somewhat by the decrease in the average propensity to consume that may 
accompany the change in distribution of money income, and by the in- 
crease in uncertainty for producers that may accompany the increase in 
the relative proportion of total demand that takes the form of producers' 
demand. 115 The net effect of all four sets of circumstances upon em- 
ployment would certainly be difficult to forecast for the economy as a 

of business cycle theory. See, for example, A. C. Pigou, Industrial Fluctuations (London, 
1927; 2nd ed., 1929), pp. 192-203, 306-313. 

114 He recognized the fact that employment might indirectly be affected favorably by 
a fall in the rate of interest as a result of a reduction in the schedule of liquidity prefer- 
ence. On the other hand, the effect on the propensity to consume of a transfer of income 
from laborers to entrepreneurs and rentiers is more likely to be unfavorable. If entrepre- 
neurs are made more optimistic by the wage change, the schedule of the marginal effi- 
ciency of capital may increase. These and other less important qualifications are con- 
sidered by Keynes, op. cit., pp. 262-267. 

115 Fellner, who is primarily responsible for propositions (3) and (4), also points out 
that both of these effects may be dampened somewhat by the decrease in output pei 
man-hour that is likely to occur. Monetary Policies and Full Employment, p. 101. 



Two lines of development are to be noted. The first was the culmina- 
tion in Hayek's The Pure Theory of Capital 116 of the discussions of the 
Austrian theory during the 'thirties. The second was the development of 
a theory of interest which comprehended monetary phenomena as having 
a direct bearing upon the determination of the rate of interest, and not 
merely responsible for temporary aberrations from the "real" situation. 

(1) The essential features of Hayek's work had been revealed in the 
course of the extended debates which preceded the appearance of The 
Pure Theory of Capital. 117 The book added relatively little to the earlier 
discussions, except that it did constitute an orderly and systematic treat- 
ment of the Austrian theory of capital and interest in the form in which 
it survived the preceding debates. In a sense the book may be said to 
belong to an earlier period than that with which the present review is 
primarily concerned, and this fact justifies not giving this very important 
book more attention. 

Hayek has been successful in developing an analysis of the capitalistic 
process for an economy which is neither stationary, on the one hand, nor 
fully dynamic, on the other. He is concerned with a sort of moving 
equilibrium for a moderately progressive economy, and is only to a minor 
extent concerned, in this volume, with the implications of his analysis for 
business cycle theory. He has avoided use of the concept of the average 
period of production and has made much less use of the idea of "stages" 
than he did in his earlier work, substituting in part input and output 
functions. His is still essentially a "time period" conception of productiv- 
ity, however, and his interest rate in an equilibrium situation is partly 
dependent upon marginal productivity, partly upon time preference. 118 
His theory is also still one which stresses the difference between the "real" 
forces as distinguished from the "more superficial monetary mecha- 



(2) The development of a theory of interest which adequately com- 
prehends the monetary elements was responsible for extraordinarily ex- 
tended and fertile discussions. 120 The primary stimulus was provided by 

118 London, 1 94 1 . 

117 The issues discussed in these debates were well summarized in: N. Kaldor, "Annual 
Survey of Economic Theory: The Recent Controversy on the Theory of Capital," 
Econometrica, July 1937, V, pp. 201-233; F« H. Knight, "On the Theory of Capital: 
In Reply to Mr. Kaldor," ibid., January 1938, VI, pp. 63-82. 

118 F. A. Hayek, "Time-Preference and Productivity: A Reconsideration," Economica, 
February 1945, XII, pp. 22-25. 

119 Idem, The Pure Theory of Capital, p. 409. 

120 In addition to Mr. Keynes' numerous and well-known contributions, the following 


Mr. Keynes' theory of interest which stressed the relation between the 
supply of money and the demand for money for cash balances purposes 
as jointly responsible for the determination of the rate of interest. Apart 
from the not inconsiderable amount of subsequent attention devoted to 
determining what Keynes really meant, the contributions stimulated by 
the liquidity preference theory may be considered under the following 
headings: (a) the formulation of a competing theory in terms of the 
supply and demand for loanable funds; 121 (b) the reconciliation of, first, 
the liquidity preference with the loanable funds approach, on the one 
hand, and second, these two monetary theories with the "real" approach, 
to the theory of interest, on the other hand; and (c) contributions that 
have improved the original formulations of the liquidity preference and 
the loanable funds theories. 

(a) The theory which regards the rate of interest as determined by 
the supply of, and demand for, loanable funds includes in the supply of 
funds: current savings, funds released from embodiment in fixed or 
working capital, and net additional bank loans; in the demand for funds: 
requirements for new investment, and requirements for a net increase in 
cash balances ("hoardings"). 122 From the very beginning, proponents of 
this type of interest theory have said that it does not lead to conclusions 
different from the Keynesian theory, although it does have the advantage 
over the Keynesian theory of corresponding somewhat more closely to the 
way in which the business world thinks of the determinants of the rate 
of interest, and it has the further advantage of showing more directly the 
relation between the marginal efficiency of investment and the rate of 
interest. Both theories, however, have the advantage over the "real" 
theories that they recognize the part played by the banking system, as 
well as the part played by the demand for speculative and precautionary 
balances, in the determination of the rate of interest. Both theories, fur- 
should be especially noted: B. Ohlin, "Some Notes on the Stockholm Theory of Savings 
and Investment," Economic Journal, March and June 1937, XLVII, pp. 53-69, 221-240; 
also in Readings in Business Cycle Theory (Philadelphia, 1944), pp. 87-130; O. Lange, 
"The Rate of Interest and the Optimum Propensity to Consume," Economica, February 
1938, V, pp. 12-32; also in Readings in Business Cycle Theory, pp. 169-192; J. R. Hicks, 
Value and Capital, Ch. 11-13, 19; D. H. Robertson, "Mr. Keynes and the Rate of Inter- 
est," Essays in Monetary Theory (London, 1940), pp. 1-38; also in Readings in the 
Theory of Income Distribution, pp. 425-460; H. M. Somers, "Monetary Policy and the 
Theory of Interest," Quarterly journal of Economics, May 1941, LV, 488-507; also in 
Readings in the Theory of Income Distribution, pp. 477-498. 

121 It should be noted, however, that such a theory had been formulated before Mr. 
Keynes' liquidity preference theory came along. Also, such a theory was employed in 
G» Haberler, Prosperity and Depression (Geneva, 1937), which appeared very shortly 
after Keynes' General Theory. 

332 If, of course, a net decrease in cash balances, rather than an increase, was to be 
expected, the item might appropriately be listed on the supply side. 


thermore, agree that the rate of saving is directly related to variations in 
the rate of interest, if at all, in a way that is not at present definable. They 
also agree that the level of saving is positively related to the level of in- 
come, the Keynesians maintaining furthermore that the marginal propen- 
sity to save probably increases as the level of income increases. 123 

(b) The loanable funds theoiy of interest can also be formulated in 
terms of the supply of, and demand for, securities and claims. II, to sim- 
plify the exposition, it is assumed that all borrowing and lending takes 
the form of the sale and purchase of a single type of security (say, a 
consol without maturity value and perfectly safe, bearing a fixed interest 
payment per period), then the rate of interest is the rate of yield of this 
security and it is determined, in any given time interval, by the supply of 
such securities coming on the market and the demand for them. The 
conditions of supply are determined by the urgency of the needs and the 
promise of the investment opportunities available to prospective borrow- 
ers, and the conditions of demand depend upon the volume of saving, the 
extent of funds released from embodiment in fixed and circulating capi- 
tal, and the lending policy of the banking system. 

It should be noted, however, that the total supply of securities poten- 
tially coming on the market must include old securities as well as new, 
since old securities may well change hands and may thus have their effect 
upon the determination of equilibrium in the market. In fact, in view of 
the relatively high proportion of existing stocks of securities to new secu- 
rities appearing on the market in any time interval, the conditions of 
supply of securities may be aflfected relatively little by the volume of flow 
of new issues. 124 

It now appears evident that the loanable funds theory, thus stated, is 
entirely reconcilable with the liquidity preference theory. According to 
the former theory, the rate of interest is the rate of yield on securities, old 
as well as new, determined by the conditions of supply and demand for 
securities in a given time interval. According to the liquidity preference 
theory, the equilibrium rate of interest is the rate at which the demand 
for cash balances will be equal to the supply of cash available for cash 

123 There is a difference between the Ohlin and the Robertson versions of the loanable 
funds theory which can be illustrated by their treatment of savings. For Robertson, the 
supply of savings in period 2 is related to the income earned in period 1 , while for Ohlin 
the supply of savings in period 2 is related to the income expected in period 2. In fact, 
there is a haziness about Ohlin's handling of ex ante savings and investment that raises 
considerable doubt as to the proper interpretation of his determinants of the rate of 

124 Cf. H. Townshend, "Liquidity-premium and the Theory of Value," Economic 
journal, March 1937, XLVII, pp. 157-158; T. de Scitovszky, "A Study of Interest and 
Capital," Economica, August 1940, VII, pp. 299-300. 


balances. The alternative to the holding of cash, for the individual, is 
conceived to be the holding of securities. Hence the theory could equally 
well be stated as follows: At the end of trading in any time interval, if 
equilibrium is achieved, the prices of securities will be such that all indi- 
viduals in the market will be content with their holdings of securities and 
their holdings of cash. Hence, if the liquidity preference theory is restated 
as a sort of "security preference" theory, it is not difficult to see that it 
involves the same determinants of the interest rate as the loanable funds 

A more formal way of showing the relation between the two theories is 
Hicks' method of counting the equations involved in an equilibrium situ- 
ation and demonstrating that either the equation for the supply and de- 
mand for claims or that for the supply and demand for money may be 
ignored, since there will be one more than the necessary number of equa- 
tions. 125 He draws the inference that, if the supply and demand for loans 
equation is retained, it is this equation which determines the rate of inter- 
est, while if the supply and demand for money equation is retained, the 
appropriate interest theory is the liquidity preference theory. 126 

Strictly speaking, the rate of interest is affected in some degree by all 
of the elements expressed in the equations in the Walras-Hicks system. 
Hence either the loanable funds approach or the liquidity preference ap- 
proach represents a form of partial equilibrium analysis of the determi- 
nants of the rate of interest. 127 The reconciliation of the two monetary 
theories and the "real" theory can best be effected, therefore, by taking a 
somewhat broader view of the economy than any one of these three 
theories ordinarily takes. 128 In the use of resources, the individual or firm 
may be conceived to be concerned with the equalization of various sorts 
of marginal rates of return : on securities, cash, production, and consump- 

125 There are n — i equations for the prices of the same number of commodities and 
factors, one equation for the supply and demand for loans, and one equation for the 
supply and demand for money. This gives usn+ i equations for the determination of n 
prices (the prices of n — i commodities and the rate of interest). Hence, one of the two 
last equations may be ignored. J. R. Hicks, "Mr. Keynes' Theory of Employment," 
Economic Journal, June 1936, XLVI, p. 246; also his Value and Capital, Ch. 12. Cf. 
also J. M. Fleming, "The Determination of the Rate of Interest," Economica, August 
1938, V, pp. 333-341- 

12,3 But the supply and demand for money equation has reference to all monetary trans- 
actions, and could be applied to the determination of the rate of interest only on the 
assumption that the prices and quantities exchanged of all commodities and factors are 
assumed given. Cf. W. Fellner and H. M. Somers, "Alternative Monetary Approaches 
to Interest Theory," Review of Economic Statistics, February 1941, XXIII, p. 44. Also, 
the supply and demand for money equation appears to have reference to a {low of trans- 
actions, rather than to the supply and demand for money to hold. 

127 Cf. W. Fellner and H. M. Somers, he. cit. 

128 The following analysis is based on H. M. Somers, op. cit. 


tion. For an individual, die marginal rate of return on securities is the 
rate of return which he can obtain by shifting additional resources into 
securities. The marginal rate of return on cash is measured by the rate 
of interest that a marginal increment of cash held might alternativelv 
earn if it were to be invested, instead of utilized for cash-balances pur- 
poses. The marginal rate of return on production is the marginal effi- 
ciency of investment. The marginal rate of return on consumption is 
equal to the rate of time preference applicable to the marginal $100 of 
income that is consumed instead or saved. 

Individuals have all four choices: to invest in securities, to hold cash, 
to invest in production, to consume. Firms have the first three choices, 
except that banks (other than the Federal Reserve Banks) have only the 
first two options. Neither the Federal Reserve nor the government acts 
on the marginal principle. Furthermore, it must be recognized that indi- 
viduals and firms probably do not apply the marginal principle rigorously. 
Nevertheless, whether individuals and firms tend to equalize all of these 
marginal returns or not, their decisions in these four areas of choice 
impinge upon the rate of interest. The supply and demand for securities, 
liquidity preference, marginal productivity, and time preference all play 
their part, then, in determining the interest rate. 

(c) Probably the most important contribution of the extended discus- 
sions stimulated by the Keynesian doctrine with respect to interest has 
been the gradual clarification of the implications of that doctrine and of 
the relation between it and other formulations of interest theory. Some 
important suggestions, however, have been made for the improvement of 
the doctrine itself. There is, first, Mr. Keynes' own suggestion, arising out 
of his debate with Ohlin, that requirements for cash balances include, in 
addition to those arising from the transactions, precautionary and specu- 
lative motives, those required in connection with current or prospective 
investment operations ("finance"). Although the original suggestion was 
rather confusing, it was gradually fitted into the structure of the liquidity 
preference theory and represents an improvement. 129 Hicks' somewhat 
broader statement that the present demand for money depends, in part, 
upon the volume of expenditure planned for the near future (including 
expenditure upon inputs and consumption, as well as upon securities) is 
a more satisfactory and inclusive statement of the amendment Mr. 
Keynes wished to make. 130 

" 9 J. M. Keynes, "Alternative Theories of the Rate of Interest," Economic journal, 
Tune I937> XLVII, pp. 246-248. Cf. E. S. Shaw, "False Issues in the Interest-Theory 
Controversy," Journal of Political Economy, December 1938, XLVI, pp. 838-856; D. H. 
Robertson, loc. cit. 

130 J. R. Hicks, Value and Capital, pp. 241-242. 


Other suggestions of considerable interest have been made by Fell- 
ner. 131 It is his position that the importance of the functional relation 
between the demand for cash balances for speculative purposes and the 
rate of interest has been exaggerated. It is not convincing, he believes, to 
argue that, in a period of depression, while other sorts of expectations are 
in general pessimistic and while the general movement of interest rates 
is downward, the expectation with respect to interest rates should never- 
theless be that they are to rise. Hence he does not accept the Keynesian 
hypothesis that the schedule of liquidity preference is highly elastic, par- 
ticularly at low rates of interest, but believes that, on the contrary, the 
schedule in question is probably quite inelastic. 132 It may even be posi- 
tively inclined. 

It follows that the shape of the schedule of liquidity preference 
(largely determined, in the Keynesian analysis, by the demand for specu- 
lative balances in relation to the rate of interest) possibly does not have 
much of a part to play in the explanation of the behavior of interest rates. 
Instead, Fellner stresses the part played by precautionary balances, the 
demand for which is conceived to shift considerably as a result of, but in 
the opposite direction to, shifts in the demand for funds for investment 
purposes. When expectations as to profits are good and the schedule of 
the marginal efficiency of capital shifts to the right, the demand schedule 
for funds for precautionary balances is likely to shift to the left, and vice 
versa. These shifts in the demand schedule for precautionary balances 
result in corresponding shifts in the schedule of liquidity preference, and 
thus affect the rate of interest. The latter is also affected, in the opposite 
way, by the change in investment. 

Regardless of whether Fellner's position concerning speculative bal- 
ances is accepted, it is quite evident that circumstances responsible for 
shifts in the schedule of liquidity preference are at least as important as 
the shape of the schedule of liquidity preference itself in the explanation 
of the behavior of interest rates. In other words, whether one adopts the 
loanable funds approach or the liquidity preference approach, the num- 
ber of circumstances that have a bearing upon the determination of the 
interest rate structure is so great that a two-dimensional, partial equilib- 
rium approach to the theory of interest must be regarded as far from sat- 
isfactory, except perhaps as a first approximation. 

3B1 W. Fellner, Monetary Policies and Full Employment, pp. 140-173. 

132 The fact that the cost of liquidity is lower at lower rates of interest is, Fellner sug- 
gests, partly or wholly offset by the fact that larger precautionary balances will be re- 
quired at higher rates of interest because of the reduction in safety margins attributable 
to the higher cost of borrowing. Ibid., p. 168. 



The theory of rent has played a very small part indeed in recent theo- 
retical discussions. The idea that rent constitutes a share in distribution 
functionally attributable to a peculiar factor, as interest is attributable to 
capital or wages to labor, is probably no longer very generally held. The 
old fires that burned so hotly have about died out. 

There is to be noted a tendency to revert to the Paretian concept of 
rent, and to define it as the surplus return which an agent of production 
earns in a particular industry over and above its opportunity cost,," 8 It is 
associated with the working of a dynamic economy, since it may result, 
for example, from an innovation, the initial effect of which is to increase 
profit for the enterprise responsible. In the course of time this profit be- 
comes shared as ''rents" with those agents of production in a position to 
bargain effectively for it. 134 Or it may result from the temporary profits 
that accompany the expansion of an industry under conditions of de- 
creasing costs, or that accompany an increase in demand in the short 
period. Under pure competition, such rents are likely to be short-lived, 
but under monopoly, oligopoly, or monopolistic competition they are 
more likely to persist. 

The concept of rent has, at least in one instance, been limited to the 
surplus return gained by an agent of production in a particular firm, over 
and above what this same agent could earn if it were employed by an- 
other firm in the same industry. 133 It would not appear likely, however, 
that rent so defined would be very common, and it is admittedly difficult 
to find illustrations that lend much importance to this version of the 


There arc essentially two kinds of profit theory: (i) the type of 
theory which regards profits as a residual, the excess of price over cost; 
and (2) the theory which regards profit as the reward for a factor of pro- 
duction, enterprise, or the entrepreneur, in the same way that wages are 
regarded as the reward for labor, and interest for capital. Both types of 

133 Cf. R. Triffin, Monopolistic Competition and General Equilibrium Theory, pp. 
I 73 _I 775 K. E. Boulding, Economic Analysis, pp. 229-232, 442-444; G. J. Stigler, Th^ 
Theory of Competitive Price (New York, 1942), p. 105; F. Machlup, "Competition, 
Pliopoly and Profit," Economica, February 1942, IX, pp. 20-21. 

VA R. Triffin, loc. cit.; B. S. Keirstead and D. H. Coore, "Dynamic Theory of Rents," 
Canadian Journal of Economics and Political Science, May 1946, XII, pp. 168-172. 

135 D. A. Worcester, Jr., "A Reconsideration of the Theory of Rent," American Eco- 
nomic Review, June 1946, XXXVI, pp. 269-277. 


theory continue to exist side by side, and there is as yet no indication of 
agreement as to the approach which is the more appropriate. 

(1) The theory which regards profits as a residual has been closely 
related, in recent times, to monopolistic competition theory. The exist- 
ence of profit may be attributed to monopolistic restriction of output, or 
to a successful "innovation" with respect to which a particular firm gets 
a head start. 13 ' 5 It may be the outcome of uncertainty, which has the 
effect of making entrepreneurs plan for wide safety margins, so that 
many firms are discouraged from entering the industry, while the more 
enterprising ones who do enter are rewarded with a profit. 137 Or it may 
arise from the fact that the industry is one in which indivisibility of im- 
portant resources constitutes a serious obstacle to freedom of entry; a 
market area, for example, may provide the existing number of firms in a 
particular industry with more than the normal profit, but may not offer 
sufficient promise of profit to justify the establishment of an additional 
enterprise in the area. 138 In all of these cases, persistence of the profit de- 
pends upon a lack of freedom of entry. Also, in all of these cases the 
profit is likely in time to be resolved, wholly or partially, into rents at- 
tributable to one or another of the factors of production. 139 

(2) Proponents of the type of theory according to which profits are 
regarded as the reward for enterprise have equally stressed the relation 
between profits and dynamic change in the economy. The attempt has 
been made, however, to relate the return to the entrepreneur as a factor 
of production and to attribute profits to some function or functions per- 
formed by him. One of the difficulties which has continued to plague this 
approach is the difficulty of defining the entrepreneurial function and of 
locating it in the corporate form of enterprise. In fact, a considerable part 
of the attention devoted to the subject of profits in recent years has been 
focused on this problem. 

Although the return to management as such is generally regarded as 
essentially a wage, functionally speaking, rather than an element in 
profits, the view is widely held that the making of those decisions as to 
policy which cannot be delegated is an entrepreneurial function. Consid- 
erable progress has been made in identifying the group within the corpo- 
ration which is usually responsible for performance of this function. 140 

186 R. Triffin, op. cit., pp. 168-179. 

137 F. Machlup, "Competition, Pliopoly and Profit," Economica, February and May 
1942, IX, pp. 15-17, 154-156. 

138 Ibid., pp. 17-19, 167-170. 

139 Cf. above, p. 45. 

1W R. A. Gordon, Business, Leadership in the Large Corporation (Washington, 1945), 
Ch. 4-1 1. 


Although outside groups or important stockholders are sometimes in- 
volved and although every corporation is probably, strictly speaking, 
unique with respect to the location of the decision-making function, 
Gordon finds that "the main elements of business leadership are exercised 
by the executive group. " 141 The reward for performance of this function 
is, however, not typically what the economist ordinarily thinks of as 
profits, but is more likely to be salary (possibly plus a bonus), combined 
perhaps with an opportunity to make capital gains by trading (with the 
benefit of inside information) in the stock of the company, and with, 
perhaps, a psychic element of prestige and a feeling of power. The execu- 
tive group may nevertheless seek to maximize the net profit of the busi- 
ness, since their efficiency and their success as executives are likely to be 
measured by the criterion of profits. If, however, the making of those 
decisions which cannot be delegated to subordinates is one of the most 
important functions of entrepreneurship, those that perform this function 
in a large corporation are mainly rewarded by forms of remuneration 
other than profits. 142 One answer, not altogether satisfactory, is that the 
firm itself as a working organization should be regarded as the entrepre- 
neur. 143 Another answer is that the compensation of certain executive offi- 
cers should be regarded as part of the profit of the corporation, from the 
economist's point of view, and should therefore not be deducted from 
gross profit in arriving at net. 144 

There remains to be considered the reward for the function of bearing 
uncertainty. It has become increasingly clear, in the discussions of uncer- 
tainty that have occurred recently, that the analysis is best to be con- 
ducted in terms of successive time periods, since the problem is essen- 
tially a dynamic one. One technique which has received considerable use 
in this connection is to express expected net returns as a probability dis- 
tribution. 145 Another technique is to conceive of each entrepreneur as de- 

141 Ibid., p. 317. 

142 J. C. Baker, "Executive Compensation Payments by Large and Small Industrial 
Companies," Quarterly Journal of Economics, May 1939, LIII, pp. 404-434; R. A. 
Gordon, "Ownership and Compensation as Incentives to Corporation Executives," ibid., 
May 1940, LIV, pp. 455-473; idem, Business Leadership in the Large Corporation, 
Ch. 14. 

143 J. H. Stauss, "The Entrepreneur: the Firm," Journal of Political Economy, June 
1944, LII, pp. 1 1 2-1 27. 

144 W. L. Crum, "Corporate Earnings on Invested Capital," Harvard Business Review, 
Spring 1938, XVI, pp. 340-341; also in Readings in the Theory of Income Distribution, 
PP- 578-580. 

115 H. Makower and J. Marschak, "Assets, Prices and Monetary Theory," Economica, 
August 1938, V, pp. 271-282; A. G. Hart, "Risk, Uncertainty, and the Unprofitability 
of Compounding Probabilities," in Studies in Mathematical Economics and Economet- 
rics (Chicago, 1942), pp. 110-118; also in Readings in the Theory of Income Distribu- 
tion, pp. 547-557; K. E. Boulding, "The Theory of the Firm," loc. cit., pp. 794-798. 


termining his policy upon the basis of expectations as to receipts, less a 
risk premium, against which is balanced the anticipated cost of invest- 
ment. 146 Entrepreneurs may be grouped according to the size of the risk 
premium they would feel it necessary to deduct from anticipated receipts 
in determining whether a particular venture would be worth under- 
taking. They also may be grouped according to their ability to bear un- 
certainty, which in turn depends upon the amount of their own capital 
and their ability to raise additional funds in an imperfect capital market. 
Thus the supply of entrepreneurial ability is limited, and there may be 
conceived to be a normal rate of profit that will compensate entrepreneurs 
for the assumption of uncertainty. When entry into an industry is con- 
templated, the anticipated rate of profit must be sufficient to cover not 
only the ordinary risk premium with respect to uncertainty as to future 
receipts and costs, but must also be sufficient to cover a risk premium 
against the possibility that other firms may simultaneously enter the in- 
dustry. Similarly, in cases in which the development of an innovation is 
involved, the risk premium must be sufficient to cover the risk that other 
firms, unknown to the first, may simultaneously be developing a similar 

In the case of the corporation, it appears then that the entrepreneurial 
function is divided between a group of the executive officers on the one 
hand, and the common stockholders (or other types of investors that 
assume a substantial burden of uncertainty as to future return) on the 
other. Whether there is any reason to expect, in a dynamic society, the 
profit which is the reward for the entrepreneurial function to tend toward 
any "normal" level is an open question. 

111 F. H. Hahn, "A Note on Profit and Uncertainty," Economica, August 1947, XIV, 
pp. 211-225. Cr. above, p. 46. 


William Fellner 

During the last ten or fifteen years, the theory of employment has pro- 
gressed significantly. It seems appropriate to characterize this advance as 
a rapid further growth of ideas that had developed up to a certain point 
in the framework of pre-Keynesian monetary and cycle theory. In the 
recent period of development a systematic theory has emerged concerning 
the relationship between certain basic functions, on the one hand, and 
aggregate output and employment, on the other, on the assumption that 
the basic functions (which, in reality, fluctuate during the cycle) arc 
'given." As compared with traditional cycle theory, this implies reduced 
emphasis on fluctuations and a more intensive and more detailed treat- 
ment of the processes occurring at any one level through which the econ- 
omy passes in the course of its fluctuations. Accomplishments along such 
lines should be fitted into a broader framework in order to become gen- 
erally applicable. In the first place, theories relating to "given" positions 
and shapes of the basic functions must become better integrated with 
theories of shifts in these functions, that is, with theories of economic 
fluctuations. Secondly, recent analysis, in its concern with relationships 
existing between aggregative concepts, has given little attention to the 
indirect effect of relative price and cost changes on movements of broad 
aggregates. This is another way of saying that the theory of employment 
must become better integrated with value theory. The really interesting 
open issues in the theory of employment bear closely on these two prob- 
lems, namely, on the relationships existing between the theory of em- 
ployment, on the one hand, and the theory of economic fluctuations and 
value theory, on the other. We shall now turn to some of these issues. 

I. The Analytical Framework: Tlte Quantity Theories vs. 
Variants of the Savings-Investment Approach 

(i) The "modern" theories of employment are rooted in monetary 
theory. The monetary theories hum which they were developed are sav- 
ings-investment theories, rather than quantity theories. This is true both 



of the "period analysis'' versions and the "equilibrium analysis" versions 
of the contemporary theories of employment. 1 They both are usually ex- 
pressed in terms of the savings-investment apparatus, rather than that of 
the quantity theories. Before contrasting the period analysis with the 
equilibrium analysis in the theory of employment, we shall raise the ques- 
tion concerning the difference in emphasis between the quantity theories, 
on the one hand, and the savings-investment framework, on the other. 
(2) Theories of employment, when developed in terms of the savings- 
investment approach, typically argue in the following way. Alternative 
amounts of aggregate money income are associated with alternative 
amounts of consumption expenditure, on the one hand, and of savings, 
on the other, income being the most important of the variables on which 
consumption and savings depend. Furthermore, income, which by defini- 
tion equals the value of the current output, is the sum of the value of 
consumption and of capital formation (provided government expendi- 
tures on currently produced goods and services are included in consump- 

1 Nothing short of bibliography filling many pages could cover the recent literature on 
these topics. We will limit ourselves at this point to the following brief references: J. M. 
Keynes, The General Theory of Employment, Interest and Money (London and New 
York, 1936); idem, "The Ex-ante Theory of Interest," Economic Journal, December 
J 937> XL VII, pp. 663-670; D. H. Robertson, Banking Policy and the Price Level 
(London, 1926); idem, "Saving and Hoarding/' Economic Journal, September 1933, 
XLIII, pp. 399-413; idem, "Survey of Modern Monetary Controversy," reprinted from 
the 1938 volume of The Manchester School in Readings in Business Cycle Theory 
(Philadelphia and Toronto, 1944), pp. 311-329; Jacob Viner, "Mr. Keynes on the 
Causes of Unemployment," Quarterly Journal of Economics, November 1936, LI, pp. 
147-167; Oscar Lange, "The Rate of Interest and the Optimum Propensity to Consume," 
reprinted from the 1938 volume of Economica in Readings in Business Cycle Theory, 
pp. 169-192; G. L. S. Shackle, Expectations, Investment and Income (London, 1938); 
M. Kalecki, Essays in the Theory of Economic Fluctuations (London, 1938); J. R. Hicks, 
Value and Capital (Oxford, 1938); Gunnar Myrdal, Monetary Equilibrium (London, 
*939); Bertil Ohlin, "Some Notes on the Stockholm Theory of Saving and Investment," 
reprinted from the 1937 volume of The Economic Journal in Readings in Business Cycle 
Theory, pp. 87-129; J. E. Meade, "A Simplified Model of Mr. Keynes' System," Review 
of Economic Studies, February 1940, VII, pp. 123-126; James W. Angell, Investment 
and Business Cycles (New York and London, 1941); Arthur W. Marget, The Theory 
of Prices (New York, 1938-42); A. C. Pigou, Employment and Equilibrium (London, 
1 941); idem, Lapses from Fidl Employment (London, 1945); Gottfried Haberler, Pros- 
perity and Depression, 3rd ed., (Geneva, 1941), Ch. 8 and 13; Alvin H. Hansen, Fiscal 
Policy and Business Cycles (New York, 1941); Mabel F. Timlin, Keynesian Economics 
(Toronto, 1942); Sir William Beveridge, Full Employment in a Free Society (London 
and New York, 1944-45); Oxford Institute of Statistics, The Economics of Full Employ- 
ment (Oxford, 1944); David McCord Wright, The Economics of Disturbance (New 
York, 1946); Walter S. Salant, "The Demand for Money and the Concept of Income 
Velocity," Journal of Political Economy, June 1941, XLIX, pp. 395-422; Clark Warbur- 
ton, "Monetary Expansion and the Inflationary Gap," American Economic Review, June 
1944, XXXIV, pp. 303-327; John H. Williams, Postwar Monetary Plans and Other 
Essays (New York, 1944), Part II; R. F. Harrod, Alvin H. Hansen, Gottfried Haberler, 
and Joseph A. Schumpeter, "Keynes' Contribution to Economics: Four Views," Review 
of Economic Statistics, November 1946, XVIII, pp. 177-196; Lawrence R. Klein, The 
Keynesian Revolution (New York, 1947). 


tion or in capital formation, which it is usual to do in the presentation of 
the theories themselves, although not in statistical practice). 2 Conse- 
quently, income is what it is becavise the consumption expenditure asso- 
ciated with that income level, plus the new capital formation, add up to 
that income. The consumption (and the saving) associated with each 
potential income level depends on the consumption function. The capital 
formation (or investment) of any period depends on profit expectations 
and on the terms on which funds are available for investment. Such a 
theory "determines" directly the amount of money income per period, 
and not the amount of real income or of employment. Certain further 
assumptions must be made concerning the behavior of wage rates and of 
prices before the conclusions relating to money income can be extended 
to physical output and employment. But the backbone of theories of this 
type consists of the savings-investment relationship. This, in turn, is de- 
pendent on the consumption function (indicating the aggregate con- 
sumption expenditures and savings forthcoming at alternative income 
levels), on some schedule expressing profit expectations for alternative 
amounts of investment, 3 and on the terms on which funds are available 
for alternative amounts of investment. 4 These functions and schedules 
are the "basic functions" on which the formal apparatus rests. 

(3) Theories of money income based on the quantity theory approach 
typically argue from the supply of money to income. Income is what it is 
because a certain amount of money is available and because this money 
is being spent at a certain rate. Traditionally, the quantity equations have 
been used more widely in the discussion of price levels than in the analy- 
sis of the determinants of income. But the "price levels" in question are 
conceived of as being determined (1) by terms, such as M and V, or as 
M and the Marshallian K, which (if multiplied or divided by one an- 
other) express money expenditures, and (2) by terms, such as T, which 
express volumes of goods purchased. Consequently, the quantity theory 
approach—even if primarily concerned with price levels— implies a way 

2 In the presentation of the theory, it would, of course, be easy to allow for the govern- 
ment expenditure on goods and services as a separate item. 

3 In the Keynesian system proper, this is the schedule of the marginal efficiency of 
capital. A. P. Lerner calls this schedule the marginal efficiency of investment (i.e., of 
the investment flow) and distinguishes it from the marginal productivity of the capital 
stock. Cf. his The Economics of Control (New York, 1944). 

4 In the Keynesian system, these terms are subsumed under the concept of "the rate 
of interest," which, in turn, is determined by the liquidity preference schedule and the 
supply of money. Consequently, in the Keynesian formal system, in which money wage 
rates are assumed as "given," the amount of employment is determined by the follow- 
ing: the liquidity preference schedule (which expresses the demand for money for 
alternative rates of interest), the supply of money, the schedule of the marginal efficiencv 
of capital, and the consumption function. All functions are expressed in terms of wage 
units, i.e., in terms of the money earnings of a labor unit. 


of looking at the determination of the size of money flows, and the in- 
come concepts are concepts of money flows. More specifically, the quan- 
tity theories imply that the size of money flows is determined by the 
available supply of money and by some expression relating money stocks 
to money flows (i.e., by M concepts, and by V or K concepts, respec- 
tively). They imply that individuals and institutions aim at some rela- 
tionship between their cash balances and their money expenditures, that 
is, at some rate of spending their cash balances. 5 

(4) The contemporary theories of employment have been developed 
mainly in terms of the savings-investment approach, rather than the 
quantity theory approach, because it is widely believed that the propen- 
sity to consume part of one's income is a truer (more "dependable") pro- 
pensity than the propensity to hold some definite amount of cash in 
relation to one's expenditures. Obviously, any completed economic proc- 
ess can be expressed just as easily with the aid of the one as with the 
other apparatus. It always turns out that the public spent some fraction 
of its income on consumption and saved the other, just as it always turns 
out that the public was spending the available cash balances at a certain 
rate. Nobody can deny the logical validity of the quantity equations. But 
this is beside the point. If, for example, the public, at the end of each 
accounting period, was merely ' 'left with" ex post cash-expenditure ratios, 
and if there existed no relationship between intentions or habits of the 
public, on the one hand, and these ratios, on the other, then the quantity 
theories would still not be logically fallacious, but they would be sterile. 
Similarly, if the propensity to consume, rather than velocity, could be in- 
dicted in this fashion, then the savings-investment approach would have 
to be regarded as sterile. Few economists would hold that cash-expendi- 
ture ratios (or velocities) actually are mere residuals in the foregoing 
sense. But many imply that they come close to being residuals, i.e., that 
they do not reflect dependable habits, while the propensity to consume 
(out of income) does. This is the reason why the savings-investment ap- 
proach has gained ground rapidly in recent times. In the framework of 
the savings-investment approach, it must, of course, be added that output 
(income) consists of investment as well as of consumption and that the 
rate of investment, which is determined by profit expectations and the 
availability of funds, is obviously influenced by changes in the supply of 

5 Active balances are being spent at some rate, and, in addition, idle balances are being 
held. The latter (tor which V == O) diminish the average rate of spending of money as a 
whole but they constitute a distinct problem. Cf. Howard S. Ellis, "Some Fundamentals 
in the Theory of Velocity," Quarterly Journal of Economics, May 1938, LII, pp. 431- 


(5) An earlier phase oi development of the savings-investment analy- 
sis may be said to have begun with Wicksell's Interest and Prices (1898) 
and to have come to a elose with the Fundamental Equations of Keynes' 
Treatise (1930). The further development and "modernization" of the 
approach during the nineteen-thirties should he attributed mainly to D. 
H. Robertson, to the neo-Wicksellians, and to the Keynes of the General 
Theory (1936). The most polemical presentation and advocacy of the 
approach is Keynes', and, at the same time, the Keynesian version has 
exerted the greatest influence on contemporary economic thought. Conse- 
quently, it has become easy to lose sight of the fact that the savings-in- 
vestment approach can be integrated easily enough with the quantity 
theory approach. One may stress the significance of the main variables 
of the savings-investment approach and yet ask the question as to the 
direct influence of the supply of money on the propensity to consume 
(i.e., on the position of the consumption function) and on profit expec- 
tations. 6 In fact, while the emphasis of Keynesians, in the narrower sense, 
may be interpreted as an "anti-quantity theory" emphasis, the reasoning 
of the advocates of the savings-investment approach has not consistently 
been directed against the quantity theories. On the contrary, some econo- 
mists who have made significant contributions to the development of the 
savings-investment analysis have also relied on the quantity equations, 
recognizing that importance may attach also to cash-expenditure ratios, as 
such, and thereby to velocities. 7 Yet the main emphasis has come to be 
placed increasingly upon savings-investment relationships, upon profit 
expectations, and upon the availability of funds. This trend of develop- 
ment has proved fruitful, and it will scarcely be reversed. The cleavage 
between this orientation and the quantity theories has recently been 
widened artificially. But this is now recognized by many economists, and 
it is unlikely that the cleavage will persist. 

(6) The Keynesian version 55 differs from some of the competing ver- 
sions of the savings-investment analysis not merely in that it presents a 
more intransigent front against the quantity theories, but in the further 
fact that the Keynesian version was developed in terms of simultaneous, 
realized magnitudes, while other versions were expressed in the frame- 
work of period analysis (sequence analysis). Economic models are estab- 
lished in such a way that they can create a presumption for dependable 
ex post relationships between simultaneous magnitudes only if either ex- 

6 In the Keynesian system proper, the supply of money affects directly merely the rate 
of interest. 

' This is generally characteristic of Professor Robertson's work. Cf. also Pigou, op. cit. 
5 The version contained in the General Theory. 


pectations are assumed to be correct or some definite assumption appears 
to be justified with respect to the nature of the errors. This is certainly 
true of the savings-investment approach. The approach, if expressed in 
terms of simultaneous, realized magnitudes, either merely states an ac- 
counting identity (in which case it possesses no significance per se) y or 
it implies that the realized equal the expected magnitudes. A mere ac- 
counting identity is stated if we say that the income of any clock-time 
period multiplied by what turns out to he the (ex post) u average propen- 
sity to consume" is the value of consumption, to which it is necessary to 
add what turns out to he the (ex post) capital formation (investment) in 
order to arrive back at the income from which we started. What turns out 
to be the aggregate saving must always equal the value of what turns out 
to be the aggregate investment. The system must always be such as to 
satisfy the ex post identity of savings and investment. 9 This accounting 
identity appears to be the main thesis of the savings-investment approach, 
as expressed in terms of simultaneous, realized magnitudes. However, if 
we postulate that an equilibrium is established in which the ex post 
(realized) magnitudes equal the ex ante (expected or planned) magni- 
tudes, then more is involved. In this event, the theory— even though it is 
expressed in ex post terms— maintains that the public decides to spend on 
consumption a definite part of its expected income and that the income of 
any planning period is what it is because the consumption so determined, 
plus the aggregate planned investment, add up to that income. So inter- 
preted, the theory is concerned with the determinants of income, on the 
assumption that a condition becomes established in which the realized 
income equals the expected, the ex post ' propensity" to consume equals 
the ex ante propensity to consume, and the realized equals the planned 

Aside from these assumptions concerning expectations, the character- 
istics of such a framework of the "Keynesian type" are those of the sav- 
ings-investment framework in general. In other words, it stresses income- 
consumption and income-saving relationships, plus the determinants of 
investment (i. e., profit expectations and the costs of obtaining funds for 
investment). A theory of this kind may be integrated with the quantity 
theory approach by investigating the relationship between the supply of 
money and the consumption function, as well as the relationship existing 
between the supply of money and investment. At any rate, such a theory 
must be supplemented by an analysis of the behavior of costs (wage-price 

9 Because aggregate income minus aggregate income times the ex post average pro- 
pensity to consume equals, by definition, the aggregate ex post savings. The same magni- 
tude also equals the aggregate ex post investment, as was just shown. 


behavior) in periods of expansion and contraction if it is to be a theory 
of real output and employment, rather than merely a theory of money 
income. All this is true of the savings-investment analysis in general and 
not merely of the particular version which runs in terms of simultaneous, 
realized magnitudes. The distinctive feature of this version is that it be- 
comes a theory of income determination only on the postulate that an 
equilibrium is established in which the realized magnitudes of the system 
equal the expected magnitudes. Otherwise it is a definitional proposition 
of no significance. 

(7) Of course, no economist suggests that, in the actual world, expec- 
tations are always fulfilled. To develop a theory on the assumption that 
they are, may serve one of two purposes. In the first place, it might be 
maintained that errors tend to cancel out in the long run. In other words, 
it might be maintained that while, for instance, the realized income of 
any single planning period is likely to be different from the income that 
was expected for that period, the deviations of ex post income from the 
ex ante (and therefore also the deviations of the ex post propensity to 
consume from the ex ante) may be disregarded in the interpretation of 
the long-run statistical relationship between income and consumption. It 
might be argued that, for the same reason, the deviations of realized in- 
vestment from planned investment 10 may be disregarded in an analysis of 
long-run statistical relationships. This is one way of "justifying" ex post 
analysis. Secondly, it is possible to maintain that it is methodologically 
convenient to separate the problem of what would happen in the absence 
of erroneous forecasts from the problem of the consequences of errors. 
Theories in terms of simultaneous, realized magnitudes would then be 
interpreted as relating merely to a hypothetical condition with correct 
forecasts, and they would have to be supplemented by an analysis of the 
consequences of incorrect expectations. For the short run, this supple- 
mentary analysis becomes necessary even if we maintain that errors tend 
to cancel out in the long run. The inclination of economists to justify 
ex post analysis in some manner such as this is largely a consequence of 
the fact that statistical data are always ex post, and therefore the use of 
statistical data for analytical purposes always implies some method of 
rationalizing ex post analysis. 

(8) This is not overlooked in the theories which were developed in 
terms of the period analysis (or sequence analysis). But in these theories, 
the equilibrium implied in the "ex post theories" appears merely as a 

10 These deviations express themselves in the involuntary (unplanned) accumulation 
or reduction of inventories, due to an excess or deficiency of actual demand as compared 
with expected demand. 


special case, and the formal framework is made suitable also for the dis- 
cussion of processes during which realized magnitudes are different from 
the expected. In the "Swedish"— neo-Wicksellian— version of the period 
analysis, the consumption of any planning period is conceived of as being 
determined by the expected income and by the ex ante (i.e., intended) 
propensity to consume. 11 Consumption— which is the product of expected 
income and the ex ante propensity to consume— and realized investment 
add up to realized income (value of output). Realized investment tends 
to differ from the planned (ex ante) investment whenever the realized 
(ex post) income differs from the expected (ex ante) income. In this 
event, the realized investment is affected by any unplanned accumula- 
tion or decrease in inventories which may occur as a consequence of the 
deficiency or the excess of demand as compared to the expected. The 
"Swedish" framework is suitable for the discussion of a dynamic model- 
process because the case in which the expected magnitudes equal the 
realized appears merely as a special case, which has been defined as mon- 
etary equilibrium. Only if realized income equals the expected, is realized 
saving equal to the expected and also realized investment equal to the 
expected. In such an equilibrium the planned (or expected) savings 
equal the planned (or expected) investment, since the realized savings 
are, by definition, always equal to the realized investment. In fact, such 
"monetary equilibrium," if it exists, may be said to be produced by the 
equality of planned savings with planned investment. Yet in the Swedish 
analysis this equilibrium is not postulated. Whenever planned savings do 
not equal planned investment, the realized magnitudes of the system will 
be different from the expected. 

However, discrepancies between expected and realized magnitudes 
make it necessary to explain expectations— and the realized data to which 
they give rise— by past experience, that is, by the realized data of earlier 
periods. If this is done, and if expectations are treated merely implicitly 
as links between "past" and "present" realized data, instead of being made 
explicit in the formal apparatus, then models of the Robertsonian variety 
are obtained. 

In the Robertsonian period analysis, the consumption of any period is 
conceived of as being determined by the income earned in the preceding 
period and by the propensity to consume out of that income. A period is 
defined in such a way that the income of any period is allocated to its use 
in the next period. Income remains unchanged if the consumption "out 
of" the income of the preceding period 12 and the aggregate investment of 

11 Which is a true propensity in the psychological sense. 
v This consumption occurs in the present period. 


the present period add up to a present income which is no different from 
the income of the preceding period. This condition is satisfied if that part 
of the previously earned income which "now" is not consumed (but is 
"saved" in the Robertsonian sense) equals the aggregate value which 
"now" is invested, so that the present consumption-plus-investment 13 
equals the income of the preceding period. If, on the other hand, aggre- 
gate investment exceeds Robertsonian savings, then income rises from 
one period to the next, and if aggregate investment falls short of Robert- 
sonian savings, then income declines. We have seen that, in the Swedish 
— neo- Wicksellian— version of the period analysis, the case in which real- 
ized magnitudes equal the expected appears as a special case (monetary 
equilibrium), and that the framework covers also the dynamic processes 
during which this condition is not satisfied. In the Robertsonian version, 
the case in which income— that is, the current value of output— remains 
unchanged is a special case. Neither the Robertsonian nor the Swedish 
period analysis implies correct expectations, although both are compatible 
with such an assumption. 

(9) It clearly would be desirable to integrate the theories of income 
and employment with the theories of economic fluctuations. These 
theories do not really pertain to different types of factual observation. 
Monetary equilibrium is not observable at all. It is a helpful concept only 
to the extent to which it is useful to interpret observable phenomena as 
deviations from such a hypothetical condition. Therefore, it cannot be 
fruitful to treat the problem of income and employment in (hypotheti- 
cal) monetary equilibrium as belonging in an area of analysis from which 
there exists no well-constructed bridge to the realistic subject of fluctu- 
ating levels of income and employment. Both the Robertsonian and the 
neo-Wicksellian period analyses contain these bridges. Analysis in terms 
of simultaneous, realized magnitudes, however, limits itself to one of the 
two areas, because it is meaningful only if monetary equilibrium is postu- 
lated. This is the real problem underlying the methodological discussion 
which for many years was apparently concerned with mere technicalities 
such as the savings-investment identity, alternative definitions of the 
multiplier, etc. 

However, the advantages of the present forms of the period analysis 
are overstated if it is maintained that those abstaining from their use 
necessarily treat the economy as if it were constantly in monetary equilib- 
rium. An economist may limit his formal— or quasi-mathematical— analy- 
sis to processes occurring "in monetary equilibrium/' and he may prefer 
to treat the problem of deviations (i.e., the problem of incorrect expecta- 

J " That is, the present earned income. 


tions) outside the formal framework, on a lower level of "generalization. " 
In this event, the formal framework may be established in terms of simul- 
taneous, realized magnitudes, but realistic conclusions cannot be derived 
before examining separately the question of the relationship between ex- 
pected and realized magnitudes— and therefore of lags— for the period 
under consideration. The reason why such a procedure may seem prefer- 
able to some economists is that the standard varieties of the period analy- 
sis are highly schematic. The Robertsonian and the Swedish period 
analyses are directly applicable only to simplified model-processes. The 
functional periods with which they operate can scarcely be defined realis- 
tically. The notion of correct and incorrect expectations implies substan- 
tial oversimplification, because expectations do not typically relate to defi- 
nite magnitudes, but to a range of probable magnitudes. Furthermore, 
the concept of expectations for the economy as a whole disregards the 
possible inconsistencies between the expectations of different individuals 
and also the inconsistencies between past and present expectations of one 
and the same individual for future dates. This specific difficulty does not 
arise in connection with the Robertsonian period analysis. The Robert- 
sonian variety of the period analysis also possesses psychological implica- 
tions, which express themselves in lags between "cause and effect," but 
the implied assumptions are comparatively simple. However, the as- 
sumed lag (between "today's" income and "tomorrow's" consumption) is 
highly schematic, that is, merely illustrative. Furthermore, the Robert- 
sonian concept of monetary equilibrium 1 * disregards the gradual growth 
in physical output which is attributable partly to the investment process 
itself and partly to exogenous factors of a rather "consistent" type (tech- 
nological progress, etc.)- A gradual increase in real output would, of 
course, have to be associated with an increase in the current value of out- 
put and of money income, unless a continuous fall in the price level is 
assumed. It is advisable to incorporate this gradual growth into the con- 
cept of dynamic equilibrium, and this requires supplementing the Rob- 
ertsonian framework, to some extent, by further elements. Robertson 
recognizes this in his discussion of policies. It is possible to include these 
as well as further considerations in the Robertsonian type of period 

Ultimately, the question is one of methodological preferences. Do we 
wish to use an analytical system of the "monetary equilibrium" variety 
(which can be defined in ex post terms) 15 and treat the problem of in- 

14 In which the value of output (income) remains unchanged by definition. 

15 That is, in terms of simultaneous, realized magnitudes, on the postulate that they 
equal the ex ante magnitudes. 


stability 10 mainly on a less formal level, without the pretension of having 
developed a generally applicable apparatus (or "system") on this second 
level? Or do we prefer to have both levels covered by the formal appara- 
tus, even though this coverage will merely "suggest" the type of approach 
to be followed in a realistic inquiry, rather than provide us with depend- 
able tools for factual analysis? 

(10) The writer's own position is influenced by the circumstance that 
economists employing the ex post systems (i.e., systems of the "monetary 
equilibrium" variety) frequently do not show sufficient awareness of the 
problems that are defined away from their technical framework. In re- 
cent years the role of several dynamic factors (capable of producing 
"cyclical" processes) was illustrated successfully with the aid of the Rob- 
ertsonian type of period analysis. 17 It is true that in most cases the analysis 
of these processes implies substantially simplified relationships between 
lagged variables as well as a great many ceteris paribus assumptions, but 
such analysis has, nevertheless, proved revealing. 

More complex and more complete "Robertsonian" systems (in an ex- 
tended sense of the word) tend to become suitable for "econometric" pur- 
poses. The merits and limitations of econometric cycle models will be 
discussed elsewhere in this volume. In these models (of, e.g., the Tin- 
bergen variety or Cowles Commission variety) the number of variables 
may become substantial and the time lags, instead of merely expressing 
schematic psychological assumptions, are chosen partly in view of the 
objective of obtaining realistic results. The ultimate question here would 
seem to be concerned with what we mean by "realistic." Systems which 
are realistic in the sense of being dependable for forecasting have not 
been invented. This is the reason why the stage at which we stop com- 
plicating the formal apparatus and start supplementing it with "judg- 
ment" has so far stayed a matter of methodological preferences. However, 
limiting the formal apparatus to monetary equilibrium analysis is an ex- 
treme procedure which easily becomes misleading. 

II. Trends and Cycles 

(i) Primary concern with the nature of the hypothetical equilibrium 
around which the economy is supposed to fluctuate is characteristic of a 

18 Instability could exist even in the monetary equilibrium of the Swedish school, be- 
cause expectations could fluctuate even i£ they were always fulfilled. But the instability 
observed in the actual world is very largely a product of monetary disequilibrium. 

17 Cf., e.g., Paul A. Samuelson, "Interactions Between the Multiplier Analysis and the 
Principle of Acceleration," reprinted from the 1939 volume of the Review of Economic 
Statistics in Readings in Business Cycle Theory (Philadelphia and Toronto, 1944), 
pp. 261-269; Lloyd A. Metzler, "The Nature and Stability of Inventory Cycles," 
Review of Economic Statistics, August 1941, XXIII, pp. 113-129. 


large portion of the recent literature in economic theory and policy. In a 
sense, one might say that theorizing in these terms takes the concept of 
monetary equilibrium at its face value. It is concerned with the develop- 
ment of the economy in monetary equilibrium, and it merely comments 
on the fact that the actual economic processes exhibit variations around 
the postulated equilibrium level. The characteristics of the assumed equi- 
librium-path are then treated as characteristics of the trend. Stagnationist 
hypotheses have mainly been developed on the basis of this type of 
reasoning. 18 

The link between the ex post ("monetary equilibrium") theories and 
the stagnation thesis may be described as follows. The assumption is 
made that in a certain kind of social and economic environment— in 
"mature" economies— the consumption function, the schedule of the 
marginal efficiency of capital, 19 and the interest rate stand in a relation- 
ship to one another which makes for a level of output that can be pro- 
duced with a considerably smaller labor input than would be required 
for full employment of the available labor force. Hence underemploy- 
ment equilibrium. The trend follows a path along which there is sub- 
stantial unemployment. There will be fluctuations around the trend in 
the course of which higher and lower levels of output and employment 
will be realized than those corresponding to the normal (or true) values 
of these variables. But what really matters is the normal level, which, in 
the circumstances here assumed, is unsatisfactory. 

The reason for this is that at satisfactory levels of activity the induce- 
ment to invest would be insufficient to absorb the substantial amount of 
savings which accrue at these levels. Income must be such that the con- 
sumption associated with that rate of income, plus the amount of invest- 
ment forthcoming, should add up to that income. On "stagnationist" 
hypotheses we would have to assume that, in mature economies, the full 
employment level of income (output) typically does not satisfy this con- 
dition. The condition is typically satisfied at a level of considerable un- 
deremployment, at which a smaller amount of aggregate investment is 
required to fill the gap between the aggregate income in question and 
the aggregate consumption forthcoming at that rate of income. This 
means assuming that, in the social and economic environments here con- 
sidered, the consumption function is too low, given the marginal effi- 

1S They were developed, aside from J. M. Keynes, op. cit., mainly in Alvin H. Hansen, 
Full Recovery or Stagnation (New York, 1938); idem, Fiscal Policy and Business Cycles; 
also in J. M. Keynes, "Some Consequences of Declining Population Growth," Eugenics 
Review, April 1937, XXX. 

19 Expressing marginal profit expectations (including interest) for alternative amounts 
of investment. 


ciency schedule and the rate of interest; or that the marginal efficiency 
schedule is too low, given the rate of interest and the consumption func- 
tion; or that the rate of interest is too high, given the marginal efficiency 
schedule and the consumption function." More precisely, the assumption 
means that these determinants of output are too low, or too high, in 
relation to one another. 

The argument concerning the position of the basic determining func- 
tions in mature economies is considered convincing by some economists 
and unconvincing by others. The writer feels unconvinced, but such a 
statement is largely subjective, because the usual argument is too broad- 
it is not specific enough— to be proved or disproved with the aid of factual 
observation. The difficulties encountered in an attempt to test these 
hypotheses may be illustrated briefly with reference to a few propositions 
frequently advanced in this connection. 

(2) The consumption function is considered "too low" 21 in mature 
economies because the full employment level of income has come to be 
associated with a high rate of physical output, 22 and because the amount 
of savings accruing at such a high rate of physical output would be too 
high to be absorbed by private investment. Since the time this view 
was first advanced, estimates have been published from which it is pos- 
sible to derive certain conclusions concerning the historical behavior of 
the propensity to consume 23 during longer periods in which physical 
output increased significantly and in which employment trends were, on 
the whole, favorable. 21 These estimates suggest that, during the last 
seventy years, the average propensity to consume has not shown a de- 
creasing trend for rising output. In the writer's opinion, this weakens 
one aspect of the stagnation thesis considerably, because it means that, 
if we should continue along the same "historical consumption function," 
investment would have to increase merely in the same proportion as that 
in which consumption rises, to fill the gap at satisfactory levels of em- 
ployment. But it must be admitted that considerations such as these do 
not really settle the issue. The absolute amount of investment required 
for filling the gap is the greater, the higher the rate of output, and it is 
possible to base pessimistic forecasts on the expectation that the absolute 
amount of investment will not rise sufficiently for full production. 

20 "Too low" or "too high" as compared with the requirements of full employment. 

21 For full employment, given the position of the other basic functions. 

22 Due to the highly developed techniques employed in advanced economies. 

23 In the ex post sense, that is, of the realized income-consumption relations for closed 

24 Simon Kuznets, National Product Since 1869 (New York, 1946). These estimates 
relate to the United States. British data seem to indicate a rising tendency for the average, 
propensity to consume. 


In fact, the stagnation thesis is concerned more specifically with the 
inducement to invest than with the consumption function (although, 
fundamentally, it is, of course, concerned with these various determi- 
nants of output in relation to one another). It is maintained that, in the 
nineteenth-century population growth, territorial expansion and capital- 
using innovations were the main stimulants of investment activity. In 
mature economies, population growth has slowed down considerably 
territorial expansion has ceased, and innovations are sometimes said to 
have lost their capital-using character. Looking at these arguments under 
a microscope, one again is apt to arrive at the conclusion that the facts 
do not lend them the support required for making a "strong case." But 
the argument is broad enough, and the facts are capable of being inter- 
preted in a sufficient number of ways, to preclude settling the issue on 
such evidence. For instance, a strong case for the population growth 
argument of the stagnationists would, presumably, require that the his- 
torical marginal propensity to consume of a community in which the 
population is growing rapidly should be greater than the per capita mar- 
ginal propensity to consume. 25 Recent estimates contradict this assump- 
tion, however, so far as the United States is concerned. 26 Historically, the 
per capita marginal propensity to consume does not appear to have been 
smaller than the marginal propensity to consume of the entire (rapidly 
growing) population. The representative family does not seem to have 
consumed a smaller proportion of its income-increments than the pop- 
ulation as a whole (which has consisted of an ever-increasing number 
of families). If a constant population could expand economically along 
a historical consumption function which is no less favorable than that 
pertaining to a growing population, why should economic expansion 
materialize more readily in conditions characterized by rapid population 
growth? It is not easy to see why this should be the case. Yet it might 
perhaps be argued that the composition of the additional output (when 
the economy expands) is more predictable if the additional output is 
produced for additional persons with similar habits than if it is produced 
for a rising per capita consumption of a given population. 27 Expansion 
materializes more readily if ceteris paribus the uncertainty attaching to 
business expectation is smaller, that is, if conditions are more predictable. 

Similar statements could be made with respect to the "territorial ex- 

25 That is, essentially, that the marginal propensity to consume of a growing popula- 
tion should exceed the marginal propensity to consume of a constant population. 

20 Cf. Simon Kuznets, op. cit., Tables II 9, II 16 and II 17. 

27 However, this argument is weakened by the fact that, even in periods of significant 
population growth, the rise in aggregate consumption reflected itself, in a considerable 
measure, in rising per capita consumption. 


pansion" argument of the stagnationists. If by "substantiating" we mean 
something in the nature of statistical testing, it would be difficult indeed 
to substantiate the hypothesis concerning an intimate connection be- 
tween territorial expansion and a satisfactory level of employment, just 
as it is difficult to substantiate a relationship between the consumption 
function and population growth. Yet the existence of frontier conditions, 
in certain segments of an economy, or colonial expansion, might prevent 
or retard the growth of institutional rigidities, and these rigidities may, 
of course, produce stagnant trends. In this connection it should, however, 
not be overlooked that the unification of political control over large 
geographical areas may share many characteristics with territorial ex- 
pansion in the narrower sense (i.e., with "conquest" in the crude sense). 
It might take another few decades before it will be possible to say whether 
the twentieth century will have brought less territorial expansion in the 
broader sense than the nineteenth. Finally, it should be pointed out 
that the "technological argument" of the stagnation thesis is also uncon- 
vincing if interpreted more or less literally. Innovation requiring less 
capital per unit of output need not for this reason provide a smaller 
stimulus to investment activity as a whole. But here, again, the growth 
of institutional rigidities may produce an environment in which the re- 
sponse to innovation is less favorable. 

In summary, it may be said that, regardless of the merits and defi- 
ciencies of the stagnationist reasoning, the Keynes-Hansen school has 
raised several significant issues which, in many respects, must be regarded 
as open issues. This is true especially if, instead of interpreting the argu- 
ments of the stagnationists narrowly, we consider the possible causal 
nexus between the phenomena they emphasize, on the one hand, and 
the degree of uncertainty and institutional rigidities, on the other. 

(3) However, at this point, we should direct our attention to the fact 
that the pessimism of the stagnationist outlook is not its only interesting 
property. It is characteristic of the reasoning that it relates primarily to 
"the trend," rather than to stages of actual dynamic development. To be 
specific, one might, for example, attempt to investigate the special cir- 
cumstances (including Federal Reserve policies and fiscal policies) that 
prevented the incipient recovery of 1931 from materializing; one might 
analyze a great many other specific factors because of which the depres- 
sion of 1929-33 turned out to be so exceptionally severe; and one might 
try to examine the after-effects of this depression on the dynamic develop- 
ment of the following decade. This certainly is one of the possible lines 
of approach, because it is obvious that during the Great Depression the 
measures which in present-day cycle theory are generally considered 


appropriate to depressions were either not adopted at all or were adopted 
on an entirely insufficient scale. Such a line of approach would not re- 
gard the trend as a meaningful phenomenon fer se y but merely as a 
convenient way of representing certain tendencies or drifts that can be 
"read from" completed dynamic processes (which alone are "real"!). It 
is characteristic of the stagnationist reasoning that it does not adopt such 
a line of approach. Instead, it argues its case primarily in terms of the 
trend itself, which is supposed to "exist" as such. It is not in specific 
phases of cyclical development that something went wrong, but the 
secular trend went wrong, and, consequently, it must be corrected more 
or less perennially by specific policies (mainly by public works, the re- 
distribution of income, and credit policies) if a quasi-chronic state of 
depression is to be avoided. The duration and severeness of the Great 
Depression and the incompleteness of the recovery during the nineteen- 
thirties are viewed as manifestations of the ailing secular trend. 

It is not accidental that pessimistic rather than optimistic hypotheses 
were expressed in terms of such a "trend theory." An economist who 
believes that with a reasonable cycle policy the general drift of economic 
development is likely to be satisfactory, is more likely to emphasize the 
cycle problem and the difficulties arising from uncontrolled cycles. We 
shall see, however, that stagnationist assumptions can also be carried over 
into cycle models, although originally they were not so expressed. 

(4) When adapted to the requirements of dynamic analysis, the sav- 
ings-investment approach is essentially a general framework for "cycle 
theory." In fact, it was originally invented to serve this purpose (Wick- 
sell), and, with a detour over the "equilibrium version," it is increasingly 
made to serve this purpose again. The detour was worth while because 
significant improvements were made on the equilibrium variety of the 
approach which can be carried over into its dynamic applications. More- 
over, the concept of monetary equilibrium has proved useful by provid- 
ing a standard with which reality may be compared. However, this tool 
also was originally invented for being used as a standard of comparison 
in dynamic theory in the framework of the savings-investment apparatus. 
Any reader of Haberler's Prosperity and Depression 28 can easily con- 
vince himself of the extent to which this apparatus has actually been used 
to develop hypotheses concerning the nature of economic fluctuations. 

When that book was first published, it was possible to group most of 
the well-known business cycle theories in two main categories: the over- 
investment theories and the underconsumption theories. The theories 
belonging in the first of these two categories emphasized the fact that 

" s Gottfried Haberler, Prosperity and Depression (Geneva, 1937; 3rd ed., 1941). 


specific scarcities may develop in the expansion phase of the business 
cycle, while other investment goods are overproduced. Investment turns 
down, in consequence of these scarcities, and, during the contraction 
phase of the cycle, it falls short of savings. One might view an over- 
investment theory as a generalized "bottleneck" theory, where the bottle- 
neck consists of the capital (or in real terms, of the specific capital goods) 
which would be required to carry the expansion further without a set- 
back. The unavailability of this capital, or of these kinds of capital goods, 
is responsible for the ensuing contraction. 29 On the other hand, the 
underconsumption theories place the emphasis on disturbances that 
might develop from the failure of aggregate consumer demand to keep 
pace with the expansion of output, i.e., from the tendency of (Robert- 
sonian) savings to outrun investment. It would, of course, always have 
been unreasonable to consider this type of theory (overinvestment or 
underconsumption theory) as anything but a simplified model or a 
standard pattern. When Haberler's book was first published, the two 
classes of theories just mentioned provided the best-known alternative 
standard patterns, although several important theories had already been 
developed which cut across these categories. 

It is undeniable that certain difficulties have existed with these stand- 
ard patterns, although their usefulness as conceptual aids is not destroyed. 
As for the overinvestment theories, downturns have occurred in periods 
which did not seem to be characterized by significant scarcities. How- 
ever, certain phenomena in the nature of scarcities could probably be 
demonstrated for any upper turning point one might select for study. 30 
Also, as Haberler argued, the economy may become gradually more 
sensitive to random disturbances as its resources become more fully 
utilized, 31 so that it is not necessary to make extreme assumptions with 
respect to the existing degree of scarcity in order to use a pattern resem- 

ay In other words, the available rate of voluntary saving becomes insufficient to con- 
tinue the expansion at an unchanging rate, and other sources of credit supply cannot be 
relied upon indefinitely because this would result in runaway inflation. The specific 
capital goods which would be required to stabilize the rate of investment at a level corre- 
sponding to the rate of voluntary saving are not available in sufficient quantities. 

30 The 1937 upper turning point in the United States would hardly be used to illus- 
trate the proposition that scarcities may put an end to cyclical expansion. Yet it could be 
argued that elements of scarcity— or "quasi-scarcities"— existed at that time. For example, 
while there existed considerable unemployment, the labor supply was restricted (the 
labor supply function had been shifted up) because of the organization of labor unions. 
Moreover, the Federal Reserve Board— in consequence of inflation fears— behaved pre- 
cisely as the Central Bank is assumed to behave in the overinvestment theories prior to 
the upper turning point. During the year preceding the downturn, reserve requirements 
were raised on three occasions. 

8i Because the expansion of any one industry in relation to economic activity as a 
whole becomes increasingly dependent on preceding actual contraction in other fields 
of activity. 


bling that of the overinvestment theories (in conjunction with other 
patterns). As for the basic pattern of the underconsumption theories, 
this raises important questions on the logical level which caused this type 
of theory to be considered an outcast in professional economics. Saving 
cannot produce contraction if it is offset by investment. Consequently, 
no underconsumption theory is logically consistent unless it provides an 
explanation of why the savings in which the ' under'-consumption ex- 
presses itself should result in hoarding instead of being offset or absorbed 
by investment. The older underconsumption theories did not come to 
grips with this problem. It is questionable how well this difficulty is 
solved in the underconsumption theories which are in vogue at present. 
(5) The pattern of the underconsumption theories can be made logi- 
cally consistent, although, of course, it never will become more than one 
of the many patterns the economist should have at his disposal when 
attempting to interpret specific economic processes. It is true that, as 
long as resources are available for expansion, the average propensity to 
consume cannot be so low or the savings ratio so high that a consistent 
willingness on the part of investors to fill the gap 32 should fail to result 
in full production. It is also true— and this is usually overlooked even 
by the modern underconsumptionists— that such 'willingness" on the 
part of the investors would justify itself from their point of view, regard- 
less of how low the average propensity to consume is, provided this "will- 
ingness" remained consistent over time. In this event, the investment 
decisions of any period would justify themselves (in the aggregate, i.e., 
aside from partial overproduction and immobility) 33 because all output 
produced would consistently be absorbed either by the consumers or by 
the investors for whom it was produced, with no involuntary accumu- 
lation of inventories. A consistently high willingness to invest is bound 
to justify itself ex fost (in the aggregate), 33 and the consumer too is 
bound to benefit from it, regardless of how low the average propensity 
to consume is, as long as the marginal propensity to consume is greater 
than zero. It certainly would be unrealistic to assume that investment 
opportunities ever were absent (or that in the predictable future they 
will be absent) in the fundamental sense of zero or negative marginal 
yields from investment, if by these "yields" we mean the returns that 
would, be realized in a full production economy in the absence of un- 
certainty, i.e., in the event that (1) the magnitude and the composition 

32 That is, to undertake enough investment to offset, or absorb, the savings. 

33 That is, aside from specific disturbances which might "generalize" themselves in 
consequence of the incomplete mobility of resources. The underconsumption theories 
are not theories of "generalized partial overproduction" in this sense. 


of consumer demand were known for each level of output, (2) inves- 
tors assumed of one another that they will always be willing to absorb 
the remaining part of the available resources for investment, and (3) 
there were no technological uncertainty. Consequently, it is a fallacy to 
maintain that aggregate output cannot be high if the average propensity 
to consume is low. It can, provided investors, in spite of the existing un- 
certainty, are consistently willing to undertake a sufficient amount of 
investment of the kind which justifies itself by subsequent further in- 
vestment demand. 34 

However, there exists a strong presumption that in an economy in 
which the average and marginal propensity to consume is low, the un- 
certainty attaching to business expectations is high. Consumer demand 
is a more stable (dependable, predictable) constituent of aggregate 
demand than is the demand for investment goods by which the gap must 
be filled. This is a consequence of the fact that consumer demand springs 
more nearly from "deep-rooted habits" of institutional (and partly, also, 
biological) character than does the demand for investment goods. Full 
production is the less likely to materialize, the higher the degree of 
uncertainty that attaches to business expectations. Even if the propensity 
to consume were very low, full production would be quite conceivable. 
But it would be unlikely, in consequence of the high degree of uncer- 
tainty existing in such circumstances. 

The increased uncertainty stems partly from the fact that investors do 
not assume of one another that they will always be willing to fill a sub- 
stantial gap by definite, predictable types of investment activity. More- 
over, if they did so, and if the gap in question were substantial enough, 
then this would imply an ever increasing ratio of real capital to real out- 
put 35 and, therefore, it would imply constantly changing methods of 
production (possibly only in the sense of 'movements along given pro- 
duction functions," but even these may be associated with a high degree 
of technological uncertainty). The writer believes that underconsump- 
tion theories can become logically consistent only if they are supple- 

34 It is generally true for expanding economies that part of the investment of any 
period justifies itself by subsequent further investment demand. This is merely another 
way of saying that the marginal propensity to consume is smaller than unity and that, 
therefore, the additional output of any period will not be completely absorbed by con- 
sumers. It must partly be absorbed by further net investment. This is true of any sequence 
of periods. It also is generally true that the investment for further investment is associ- 
ated with investment for additional consumption. This is merely another way of saying 
that the marginal propensity to consume, while smaller than unity, is positive. Cf. p. 77, 

35 Cf. the present writer's Monetary Policies and Full Employment (Berkeley, 1946), 
pp. 43-46; E. D. Domar, "Expansion and Employment," American Economic Review, 
March 1947, XXXVII, pp. 34-55. 


mented by an appropriate theory of uncertainty. Otherwise, these theories 
would either imply zero marginal efficiency— that is, lack of investment 
opportunities— in the "fundamental" sense (' net-of-uncertainty" sense) 
discussed on p. 66; or they would imply a very low marginal efficiency 
in this fundamental sense plus a floor to interest rates set by infinitely 
elastic liquidity preference. While Lord Keynes made the assumption 
(which seems implausible to the present writer) that these factors might 
become operative in the predictable future, it would be clearly unrealistic 
to assume that in the industrialized economies they have been operative 
because historically these economies have not experienced diminishing 
returns. Cycle theories other than underconsumption theories do not have 
to overcome this conceptual difficulty because they typically argue from 
specific disturbances (partial overproduction^) to general contraction, with 
reference to the incomplete mobility of resources. 36 The underconsump- 
tion theories, however, are not theories of generalized overproduction in 
this sense. They argue directly in over-all terms, not via partial overpro- 
duction and immobility. Consequently, they have to explain why there 
should be an over-all deficiency of inducements to invest. 

(6) Neither the older nor the contemporary versions of the under- 
consumption theory bring out these points clearly, but some of the con- 
temporary versions can be translated into these terms. They then become 
logically consistent, which, of course, does not of itself settle the question 
of their contribution to the understanding of economic fluctuations. The 
"modern" versions of the underconsumption theory are essentially the 
dynamic versions of the stagnation thesis. They argue that, in the cir- 
cumstances existing in "mature" economies, investment becomes insuffi- 
cient to fill a very substantial gap. In consequence of the slowing down 
of extensive growth (population growth, territorial expansion), 37 the in- 
ducement to invest is too weak to accomplish this. Consequently, in the 
course of the expansion, a point is reached at which the gap between 
aggregate output and consumption becomes too great to be filled by the 
amount of investment which is forthcoming in the present social-eco- 
nomic environment. The economy cannot get beyond such a critical 
point, which may be one of substantial underutilization. This, however, 
does not answer the question of why the necessary amount of investment 
is not undertaken, even though (if it were undertaken consistently over 

30 Partial overproduction may give rise to a deflationary spiral and to generalized over- 
production if resources cannot be shifted freely from the original fields of overproduc- 
tion. Practically all cycle theories, other than the underconsumption theories, are ulti- 
mately based on this notion, provided they claim to be general cycle theories. 

37 Some would add: also in consequence of the changing character of technological 


time) it would turn out to be profitable, regardless of population growth, 
territorial expansion, and the numerical value of the propensity to con- 
sume. Yet it may be possible to argue that, in the course of cyclical 
expansions, the uncertainty attaching to business expectations increases, 
due to the fact that the proportion of output absorbed by consumer de- 
mand falls and the proportion absorbed by investment activity rises. It 
may also be possible to argue that the slowing down of extensive growth 
has contributed to the development of institutional rigidities by which 
the uncertainty has been increased. Obviously the fact that a sufficient 
amount of investment (if it were undertaken consistently over time) 
would justify itself in the aggregate is compatible with any degree of 
uncertainty from the viewpoint of the individual investor. 

(7) We may conclude that the standard patterns of the undercon- 
sumption theories and of the overinvestment theories can be made logi- 
cally consistent, and that they continue to belong among the models the 
economist should have at his disposal when attempting to interpret 
dynamic processes. Other significant models cut across the distinction 
between the two older theories just discussed. During the last decade, 
some of these were developed further, and, on the whole, more emphasis 
has been placed on these patterns than on the two models so far con- 
sidered. Even those writers who come nearest to "representing" modern- 
ized versions of one of the two traditional theories have combined these 
with elements taken from further patterns (e.g., Hansen). 38 

Of these further models, the innovation theory should be mentioned 
first. This conceives of dynamic development as being produced by waves 
of innovations (i.e., by waves of "setting up new production functions") 
and by the adaptation of the economic system to the new methods of 
production (Schumpeter). 39 Long waves of economic development— 
"Kondratieff" upgrades and downgrades— are interpreted as having been 
produced by waves of innovations, predominantly of a definite type (or 
of a limited number of types) by which the specific long wave in question 
is characterized. The Kondratieff wave (upgrade plus downgrade) 
stretching from the seventeen-eighties to 1842 is that of the industrial 
revolution; the next (1842-97) is that of steam and steel; the third 
(from equilibrium in 1898 through an upper turning point prior to 
the first war and a trough in the 'thirties, toward future equilibrium) 

3S Cf . mainly Alvin H. Hansen, Full Recovery or Stagnation; idem, Economic Policy 
and Full Employment (New York and London, 1947). 

38 Cf. mainly Joseph A. Schumpeter, Business Cycles (New York and London, 1939); 
for a brief outline of the theory, cf. idem, "The Analysis of Economic Change," reprinted 
from the 1935 volume of the Review of Economic Statistics in Readings in Business 
Cycle Theory, pp. 1-19. 


is the long wave of electricity, chemistry, and motors. Shorter waves 
— Juglar cycles and Kitchin cycles— become superimposed upon these 
long waves, as innovations spread, and as they give rise to related inno- 
vations, gradually changing the character of the production processes 
throughout the economy. The Juglars appear to be one-sixth as long as 
the Kondratieffs and three times as long as the Kitchins. 40 The entire 
process is started by entrepreneurs (innovators) of exceptional talent, 
and it propagates itself through imitation by others. The scope of the 
innovation theory of economic development is sufficiently broad to have 
rendered it possible for its author to discuss modern business cycle history 
from the angle of a unique central hypothesis. 

(8) Other important patterns, contributing to the explanation of 
dynamic processes, are more specific in the sense of relating to narrower 
aspects of these processes. This, perhaps, is the least true of the Acceler- 
ation Principle, which, while concerned with a specific aspect of dynamic 
development, nevertheless lends itself more readily to being extended in 
scope than do some of the other patterns. The most significant proposition 
involved in the Acceleration Principle is that a decreasing rate of growth 
during the expansion tends to lead indirectly to an absolute decline of 
business activity. The durable goods industries, and also the production 
activities resulting in goods "for inventories," are more nearly geared 
to some definite rate of growth than to some definite level of output. 
If, after a period of rapid growth, the economy tends to become tempo- 
rarily stabilized at a higher level, then the output of the specific industries 
in question will fall to the mere replacement rate. 41 Therefore, the 
primary tendency of the economy toward stabilization (or toward a 
slower rate of expansion) might not actually result in stabilization (or in 
a slower rate of expansion) but might, instead, result in a significant 
contraction of output. Since the basic underlying forces— such as dis- 
coveries, population growth, etc.— would, of themselves, tend to produce 
growth at varying rates, it is possible to arrive at models of economic 
fluctuation, assuming that the response of the economy to these basic 
forces will be influenced by the Acceleration Principle. 42 

40 For the three cycles thus resulting, cf. note 43, p. 71. 

41 Of the stock of durable goods and of inventories. 

42 Cf. J. M. Clark, Strategic Factors in Business Cycles (New York, 1935)- It should 
be added that more recently the Multiplier theory and the Acceleration Principle have 
also been synthesized into cycle models (R. F. Harrod, The Trade Cycle [Oxford, 1936]; 
Paul A. Samuelson, op. citS). The underlying idea here is that additions to investment 
give rise to subsequent additions to income at a decreasing rate, due to the multiplier, 
and that such a primary tendency toward change at decreasing rate may generate fluctu- 
ations via the Acceleration Principle. However, it should not be overlooked that the 
Multiplier-Acceleration process operates in the fashion described only if the original 
additions to investment result merely in subsequent consumption spending, except for 


(9) Some of the further cycle patterns can only be mentioned briefly. 
The "cobweb" pattern was further developed, and also subjected to 
criticism, in the course of the last decade. The pattern describes fluctu- 
ations which, on rather specific assumptions of the theorem, are induced 
by a consistent tendency on the part of producers to shoot beyond the 
mark, i.e., by excessive increases in output when prices are high and by 
excessive curtailments of output when prices are low. Each individual 
producer is inclined to believe that prices will stay high (or low) until 
he completes his adjustment to the prices "now" prevailing, but since 
a great many producers act simultaneously, prices will have turned from 
high to low (or from low to high) by the time the individual producers 
have completed the expansion of their production, or its contraction, as 
the case may be. Certain cycles in agricultural production (corn-hog 
cycle) may be partly explained by this model. 

Building cycles, which have a duration of between fifteen and twenty 
years, may also have something to do with processes of this kind, although 
their duration excludes placing the emphasis on the cobweb model alone. 
The reinvestment cycle model has also been used to explain waves in 
building activity and in the production of other durable assets. Here 
again the mere fact that durable assets must be replaced after some time 
does not explain the occurrence of cyclical fluctuations, because replace- 
ment dates are by no means uniquely determined by the physical con- 
dition of capital goods. But, since replacement is not indefinitely 
postponable, the reinvestment-cycle model may contribute to the under- 
standing of fluctuations in the production of durable goods (such as 
houses), and it may contribute even to the explanation of economic 
instability in general. 

The building cycle has been emphasized considerably in the recent 
literature on dynamic processes. The average duration of this cycle is 
much greater than that of the "business cycle" proper, 43 but building 
cycle turning points (especially the upper turning points) tend to show 

such further investment as is induced by this additional consumption expenditure. If the 
original additions to investment result directly in additions to (or reductions in) other 
kinds of investment activity— aside from the investment induced by the consumption 
effect of the original investment— then the process becomes more complicated, and it 
should be approached in terms of the marginal propensity to spend (James W. Angell, 
of. cit.} rather than in terms of the marginal propensity to consume. Metzler's model 
of inventory cycles is also based on an ingenious blending of the Multiplier theory with 
what may be considered a variant of the Acceleration Principle (cf. Lloyd A. Metzler, 
of. cit."). 

43 In the United States, roughly five times as long, if by the business cycle we mean 
the reference cycle of the National Bureau of Economic Research. This cycle is defined 
as expressing fluctuations in aggregate economic activity, with a duration of more than 
1 year but no more than 10-12 years, and not divisible into shorter cycles with ampli- 
tudes approximating their own. The average duration of these for longer periods is 


a lead in comparison with the corresponding (i.e., nearest) business cycle 
turning points. Furthermore, the most seriously depressed periods of the 
last hundred years or more fall in downgrades of the building cycle. 44 
They also fall in longer intervals marked by a downward trend in prices 
and interest rates, that is, they fall in the "KondratiefT downgrades" 
(long-wave downgrades) of Schumpeters three-cycle model. The direc- 
tion of causation can, of course, not be decided by statements such as 

(10) The various theories surveyed express possible contributing 
causes and illustrate them with simplified models. They all may be said 
to fit into the doctrine of generalized partial overproduction, 45 except 
insofar as they have underconsumptionist implications. Ultimately, gen- 
eral contraction is explained either by partial overproduction plus immo- 
bility or by over-all underconsumption, although both basic types of 
explanation include many variants. Some specific theories (e.g., the 
overinvestment theories) are clearly variants of one of these two basic 
types. Others (e.g., the Acceleration Principle) may be fitted into either 
type. Which of these one wants to use— or to emphasize more than the 
alternative ones— in the discussion of a given period depends on the spe- 
cial characteristics of that period. This is not true of the savings-invest- 
ment framework*" which is suitable for expressing any specific theory 
on the nature of dynamic processes. But the statement is true of the 
specific theories of which the body of "business cycle theory" mostly 
consists at present. Each of these expresses something in pure form that 

slightly in excess of 40 months for the United States. The reference cycle corresponds 
roughly to the Kitchin cycle of Schumpeter's schema in that it relates to business in gen- 
eral (as all three cycles of the Schumpeter schema) and has an average duration of close 
to 40 months. The dating is different because the "Kitchin" is founded on the hypothesis 
that two longer cycles are superimposed upon it, while the National Bureau identifies 
the turning points of its reference cycle from the ups-and-downs of time series directly. 
Moreover, Schumpeter dates his cycles from "equilibrium neighborhood" through peak 
and trough to next equilibrium, while the National Bureau dates from trough to trough 
with no equilibrium implications. The three-cycle hypothesis relates to fluctuations in 
aggregate economic activity, although the KondratiefT wave is much better established for 
prices and interest rates than for physical time series. The building cycle discussed in the 
text relates directly to a specific kind of activity. For Kondratieff's findings, cf. Nikolai D. 
KondratiefT, "The Long Waves in Economic Life," reprinted from the 1935 volume of 
the Review of Economic Statistics in Readings in Business Cycle Theory, pp. 20-42. 
For criticism, cf. George Garvy, "KondratiefT's Theory of Long Cycles," Review of Eco- 
nomic Statistics, November 1943, XXV, pp. 203—220. 

44 The eighteen-seventies, the eighteen-nineties, and the Great Depression. 

45 Cf. p. 68, including note 36. 

46 That is, the framework based on the proposition that income rises if investment 
exceeds Robertsonian savings and that jt falls if Robertsonian savings exceed investment; 
or on the proposition that expectations turn out to have been too cautious if investment 
ex ante exceeds savings ex ante, and that they turn out to have been insufficiently cau- 
tious if savings ex ante exceed investment ex ante. 


in reality is found at best in highly impure form and, in certain periods, 
may not be found at all. Further progress depends, therefore, on the 
co-ordination of analytical thinking with empirical research. 

(i i ) Empirical research has recently mainly proceeded along two lines. 
The first is characterized by the attempt to formulate systems of equations 
with a great many (dated) variables and to test how well the observed 
values of these variables fit alternative systems during some period ol 
time. The problems raised by this method are mainly problems in statis- 
tical theory, and they will not be discussed here. An appraisal of the 
method would have to throw light on the question of what is proved by 
showing that good fits are obtained with certain systems. 47 However, it 
should not be overlooked that the much less sophisticated "garden vari- 
ety" of "testing" hypotheses for plausibility (which means using few 
variables and not applying refined statistical tests) is subject to the same 
limitations, in addition to operating with obviously incomplete "systems." 
The question is essentially that of finding the appropriate limits to refin- 
ing a method which, given the character of the available material, is 
incapable of contributing more than a certain amount to the understand- 
ing of a problem. 

The other empirical line of approach is that followed by the National 
Bureau of Economic Research (Mitchell, Burns, Mills, and others). 48 
This method rests on the observation of the behavior of a great many 
specific time series during their own "specific cycles" and during the 
'reference cycles." Specific cycle turning points are those of the specific 
time series in question, while reference cycle turning points are those of ag- 
gregate economic activity. 49 After the elimination of seasonal variations, 
the values assumed by each time series are plotted for nine stages of each 
specific cycle and for the nine corresponding stages of each correspond- 
ing reference cycle (in terms of index-numbers, with the average value 
of the time series during the cycle in question as the base). In this fash- 
ion, a specific cycle pattern and a reference cycle pattern is obtained 
for each time series, for each cycle. Subsequently, both the specific cycle 
and the reference cycle patterns are averaged for longer periods, that is, 

47 Considering the very substantial margins of error to which the observations are sub- 
ject, and considering the sensitiveness of these systems to changes in their values, the 
question also arises as to what is disproved by showing that certain systems do not give 
good fits. 

** Cf. Arthur F. Burns and Wesley C. Mitchell, Measuring Business Cycles (New 
York, 1946); Frederick C. Mills, Price-Quantity Interactions in Business Cycles (New 
York, 1946). 

49 In setting the reference cycle turning points, the National Bureau takes into account 
the information obtainable from Business Annals (Thorp) concerning the state of busi- 
ness in general, and also the clustering of specific cycle turning points around certain 
dates (cf. also note 43, p. 71). 


the value found for each of the nine stages of the individual cycles is 
averaged over all cycles 50 which developed during the period. By this 
procedure, an "average" pattern is obtained and plotted for each time 
series, expressing its "average behavior" during its own specific cycle, 
on the one hand, and during the reference cycle, on the other. 51 The 
method lends itself to studying many significant problems, such as those 
relating to the duration and the amplitude of the cycles observed in the 
various time series; leads and lags of various time series as compared to 
others and as compared to business in general (i.e., to the reference 
cycle); the possible existence of higher cyclical units consisting of triplets 
of reference cycles 52 or of all reference cycles lying between certain par- 
ticularly severe depressions; the relative intensity of price changes and 
changes of output in the various stages of the cycle, etc. 

The National Bureau technique is different in several respects from 
the traditional techniques by which "cycles" have been found, i.e., the 
raw material for cycle research has been obtained. Materials so obtained 
have been used for many purposes, ranging from the mere graphic repre- 
sentation of the cyclical behavior of time series to the testing of the sys- 
tems of equations previously mentioned. The following appear to be the 
most significant differences between the traditional techniques of decom- 
posing time series and the National Bureau method. In the first place, 
customarily the secular trend has been eliminated prior to identifying the 
cycle. This means that the cycle expresses fluctuations around the trend. 
It follows from the description on p. 73 that the National Bureau first 
establishes the turning points of the cycle, and afterwards eliminates 
merely the inter-cycle portion of the trend (by defining each cycle as a 
movement around the average value during the cycle itself^). The intra- 
cycle portion of the trend is not eliminated. Furthermore, traditionally 
merely "specific cycles" were found, although some of the time series for 
which they were established are of general significance as indicators of 
business conditions. Application of the concept of the reference cycle to 
specific time series is a characteristic feature of the National Bureau 
method. The writer regards both these new features as improvements. 
They express an outlook characterized by the postulate that the aggregate 
of the economic developments between any reference cycle trough and 
the successive trough constitutes a "real" or "meaningful" experience in 
the life of a social community. The averaging of the cycle patterns, for 
each time series over longer periods, which also is an original feature of 

w Over the specific cycles, on the one hand, and the reference cycles, on the other. 
61 "Average behavior" during a longer period comprising many cycles. 
52 That is, Schumpeter's Juglar cycle (cf. note 43, p. 71). 


the National Bureau method, expresses the idea that all these pieces of 
experience belong in a common family of experiences. 

The main limitations of the method are connected with the use of 
averages as representative values. But, of course, some difficulty of this 
character must necessarily arise in any attempt at generalization. The Na- 
tional Bureau method makes it comparatively easy for the reader to gain 
an impression of how typical (or untypical, as the case may be) the 
average pattern of a time series was of its changing behavior during a 
longer interval of time. 

III. Rigidities and the Problem of Uncertainty 

(i) Bridging the gap between recent developments in the theory of 
employment, on the one hand, and value theory, on the other, will prob- 
ably prove a difficult task. The modern theory of employment lives a dis- 
concertingly independent life from value theory. The artificial separation 
of these two fields expresses itself most clearly in the fact that the modern 
theory of employment has little to say about the effects of changes in the 
cost-structure on aggregate output and employment. This gives the theorv 
an essentially monetary character, even if the analysis appears to run in 
real terms. The 'real" magnitudes are derived on drastic simplifying as- 
sumptions with respect to the price and wage level. 

(2) One of these drastic assumptions is that a change in the general 
level of money wages does not affect the level of aggregate output and 
employment, except through its repercussions via the interest rate (i.e., 
except by changing the existing degree of liquidity in relation to the 
money requirements set by the volume of transactions). This latter quali- 
fication—with respect to repercussions via the interest rate— does not 
relate to any really distinctive effects of wage changes, because the effect 
via interest rates is precisely identical with that of changes in the supply 
of money, given all prices and wages. 53 The proposition concerning the 
neutrality of general wage changes 54 is derived from the assumption that 
prices tend to change in the same direction and in the same proportion as 
money wage rates. At present, this proposition is perhaps rarely main- 
tained in quite the crude and definite form here presented. But it fre- 
quently is maintained in approximately this form. 

The best way of understanding the limitations of the general-wage- 

63 In other words, the effect in question is a "purely monetary" effect. Lord Keynes 
tended to hold this view, although the view is qualified in Chapter 19 of the General 

61 Except for the "purely monetary" effect just mentioned. 


level neutrality idea is that of inquiring into the conditions under which 
it would be true. This seems preferable to repeating the line of argument 
along which its justification is usually presented. 55 The proposition would 
be true if it could be taken for granted that, at each level of real output, 
a given rate of consumption per period of time tends to "justify"— and, 
therefore, to bring about— a definite rate of new investment per period. 56 
If this were taken for granted, it actually would be inconsistent to argue 
that a change in money wage rates produces a change in real wage rates 
and, thereby, a change in real output. For, with the change in real wage 
rates, the amount of consumption would have to be different, at all levels 
of output, from the consumption that would have materialized prior to 
the wage change. For example, if we envisage a money wage reduction 
that results in reduced real wage rates, we must conclude that consump- 
tion at any level of output will tend to be smaller than would have been 
the case without the wage reduction. 57 Consequently, at any level of 
physical output, a different amount of investment (in. the case of a wage 
reduction, a greater amount of investment) 58 would now be forthcom- 
ing per unit of simultaneous consumer demand. If it is postulated that 
this cannot be true— i.e., that at each level of output a given amount of 
consumption "justifies" no more and no less than a definite rate of new 
investment— then the change in money wage rates must not be assumed 
to shift the consumption function. 59 Hence, the change must not be as- 
sumed to result in changing real wages. Prices must be assumed to change 
in the same proportion. 

Yet, if producers, under the influence of changing money wage rates, 
change the amount of net investment, per period of time, which is asso- 
ciated with a given amount of consumption (i.e., with a unit of simulta- 
neous consumer demand), then the outcome is different. For example, if 
we postulate that a decrease in money wage rates induces businessmen to 
undertake more investment "per unit of consumer demand," then real 
wage rates will decrease. 60 In this event, total income (consumption plus 

55 This line of argument rests ultimately on the notion that in the very short run, wage 
costs are the only prime costs, and that, consequently, in the short run, prices adjust 
completely to the change in money wage rates. This means that reproduction costs also 
adjust, and that, consequently, there is no reason tor a different outcome in the long run. 

56 Assuming that interest rates do not change, aside from the "purely monetary" effect 
(repercussions via the interest rate) considered in the preceding paragraph. 

57 Because the propensity to consume out of wage income is typically greater than out 
of non-wage income. 

58 In the event of wage increases: a smaller amount of investment. 

5y Because the ratio of consumption to investment must not be assumed to shift for any 
level of output. 

,X) If we assume that less investment is undertaken in relation to consumers' demand 
then real wage rates will rise. 


investment) will become greater 01 in relation to consumer demand and, 
therefore, it will also tend to become greater in relation to the wage bill. 6 " 
Non-wage income will become greater in relation to wage income, which 
is another way of saving that (given the man-hour output) real wage 
rates will decline. Prices will fall in a smaller proportion than money 
wage rates hecause aggregate demand (consumption fins investment) 
will fall in a smaller proportion than consumer demand. Postulating that 
investment increases per unit of consumer demand means postulating 
that a condition will establish itself in which the average propensity to 
consume is lower than was the case before. Such a condition will estab- 
lish itself by a fall in real wage rates. 

It seems, therefore, that the crucial question in this connection relates 
to the amount of investment which appears to be "justified" to producers 
per unit of consumer demand. There exists an underconsumptionist bias 
to the effect that this amount of investment tends to be a constant, aside 
from temporary fluctuations around a "normal" value. Or perhaps the 
implication is that the "normal" (and "justified") amount of investment 
is determined by the marginal propensity to consume— i.e., by the slope 
of the consumption function— because this determines the increase in 
consumer demand in which the investment and the attending rise in out- 
put results. In fact, as we have seen, the proposition that changes in the 
general level money wage rates produce no effect on output is justified 
only on the assumption that the ratio of new investment to consumption 
is a constant (or that it tends to be a constant, aside from erratic fluctua- 
tions around its "normal" value). 

(3) There exists no such simple relationship between the "justified" 
amount of investment, on the one hand, and the simultaneous rate of 
consumption, 63 on the other. Widely different amounts of aggregate in- 
vestment may prove to be justified per unit of simultaneous consumption, 
provided the willingness to invest stays high in the long run, so that, in 
each subsequent period, the justified amount of aggregate investment 
again is considered to be high, per unit of consumption. The additional 
output to which the new investment gives rise is subsequently always 
partly absorbed by consumers and partly by further (subsequent) inves- 
tors. (This is always true if the marginal propensity to consume is smaller 
than unity.) If subsequent investors are consistently willing to absorb 
enough of the additional output to which previous investments gave rise, 

01 In the event of a wage increase : smaller. 

C2 Considering that the propensity to consume out of wage income is greater than out 
of non-wage income. 

68 Or the marginal propensity to consume. 


then the economy may go on expanding indefinitely, regardless of how 
low the propensity to consume is. 64 Of course, there always will be some 
rise in consumption, as well as in investment, as long as the marginal pro- 
pensity to consume is greater than zero. No logical inconsistency is in- 
volved in visualizing an expanding full-employment economy in which 
output consists of consumption to the extent of i o per cent and of invest- 
ment to the extent of 90 per cent, just as there is no inconsistency in the 
more realistic assumption that consumption accounts for about 90 per 
cent and investment for about 1 o per cent. In both economies, the addi- 
tional output, which is attributable to the investment, is partly absorbed 
by subsequent consumption and partly by subsequent investment, and 
this is true of any sequence of periods. But in the first of these two econo- 
mies, a much greater part is absorbed by subsequent further investment. 
This is a very significant difference, affecting importantly, among other 
things, the degree of uncertainty existing in the economy. 65 But within 
reasonable limits, the proportion of output which is absorbed by invest- 
ment is actually variable (not merely "in principle," i.e., not merely if we 
disregard uncertainty). There is no reason to assume that the "justified" 
amount of investment should bear some simple relationship to the simul- 
taneous rate of consumption or to the marginal propensity to consume. 
It seems quite likely that a change in the general level of money wages 
would actually tend to change the ratio of investment to consumption per 
period of time; that is to say, it is quite likely that investment would not 
be changed in the same proportion as consumption. When we are con- 
cerned with consumption output, it may be reasonable to assume, in the 
first approximation, that the influence of wage changes is equally signifi- 
cant on cost, on the one hand, and on demand, on the other. With respect 
to investment, as a whole, such a premise does not seem equally plausible. 
For such investment activity as results in goods to be absorbed by subse- 
quent, further investments, 66 wages are direct cost-factors, but they are not 
demand-influencing factors in any direct way. If it is assumed that general 
money wage increases reduce the ratio of new investment to consumption 
and that general money wage reductions raise the ratio of new investment 
to consumption, then money wage increases will tend to be associated 
with increased real wage rates and money wage reductions with reduced 

64 Unless the marginal efficiency of capital, in the "fundamental" or "classical" sense 
discussed on pp. 66-68, declines to zero or to the institutional floor level of interest 

05 It affects uncertainty in more than one way. Cf . p. 67, above. 

06 Which, in turn, again result in goods to be absorbed by subsequent, further invest- 
ment, etc. 


real wage rates. In this event, aggregate demand adjusts incompletely to 
the wage changes, because investment demand adjusts incompletely. 
Therefore, prices also adjust incompletely. We are implying here that 
after the change in wage rates, wages are expected to stay at their new 

There is much room for theoretical as well as empirical investigation 
on this subject. While, on the whole, there still exists a tendency to get 
around these problems with drastic simplifying assumptions, some work 
has been done on the relationship between changes in money wage rates, 
on the one hand, and changes in real wage rates, on the other (Dunlop, 
Tarshis). 67 It is likely that, in the future, we will witness a renewal of the 
effort to formulate a reasonably realistic theory of the relationship be- 
tween money and real wages. It also seems likely that the effect of wage 
changes on aggregate output and employment will be investigated fur- 

(4) So far as the relationship between changes in real wage rates and 
employment is concerned, much of the theoretical thinking of recent 
times is either based on highly "orthodox' assumptions or on a simple 
reversal of these. To get the problem already discussed out of our way, let 
us assume that we are able to trace the effects of money wage changes on 
real wages, and that we are faced with a given change in money wage 
rates and with an attending change in the general level of real wage rates. 
Should it be taken for granted that aggregate output and employment 
will always move inversely to real wage rates? Or that it will move in the 
same direction, due to a favorable effect of wage increases on effective 
demand? Recent theory has had little to say on the subject, although 
common sense reasoning should show clearly enough that neither of 
these categoric views can possess general validity. If real wage rates rise 
beyond certain limits, the reward for bearing uncertainty must become 
insufficient to induce an adequate amount of investment activity. More- 
over, there will develop a tendency to substitute capital for labor. If real 
wage rates decline to very low levels, the propensity to consume also de- 
clines very low, and production and employment could be high only if an 
unusually high proportion of output were allocated to further investment 

67 John T. Dunlop, "The Movement of Real and Money Wage Rates/' Economic 
Journal, September 1938, XLVIII, pp. 413-434; Lorie Tarshis, "Changes in Real and 
Money Wages," reprinted from the 1939 volume of the Economic Journal in Readings 
in the Theory of Income Distribution (Philadelphia and Toronto, 1946), pp. 330-335; 
J. M. Keynes, "Relative Movements of Real Wages and Output," Economic Journal, 
March 1939, XLIX, pp. 34-52; cf. also John T. Dunlop, Wage Determination under 
Trade Unions (New York, 1947); Lloyd G. Reynolds, "Wage Differences in Local Labor 
Markets," American Economic Review, June 1946, XXXVI, pp. 366-375. 


(rather than to consumption) in each subsequent period. As was pointed 
out before, such a condition is logically quite conceivable; most of the 
investment of each period could result in output for further investment in 
the subsequent period. If the willingness to invest should— in the long 
run— not slacken in these circumstances, the investments would prove 
profitable. Aggregate consumption would also be rising, provided the 
marginal propensity to consume was greater than zero (which, of course, 
should be assumed). But while such a condition is quite conceivable, it 
would create a very high degree of uncertainty because the system cotdd 
cease to function satisfactorily whenever the willingness to invest for fur- 
ther investment declined. Owing to this uncertainty, the system would 
presumably never start functioning satisfactorily in conditions such as 
these. Consequently, "too high" real wage rates and "too low" real wage 
rates seem equally incompatible with full production. This raises the 
problem of an optimum, which so far has received insufficient attention. 

(5) On the whole, the framework in which much of the recent analy- 
sis was developed is not particularly well suited for coping with tasks 
which involve the problem of uncertainty. Most analytical systems "get 
rid" of the uncertainty problem by a procedure which easily becomes mis- 
leading. 08 Marginal revenue is equated to marginal cost, or the marginal 
efficiency of capital is equated to the rate of interest! These, of course, 
are merely variations on the profit-maximization theme. 69 But what kind 
of "expectations" are expressed in these functions? For, surely, the pro- 
ducer should be said to equate the expected values of these functions. 
The implication of the conventional procedure is that the functions in 
question express something in the nature of mathematical expectations 
(or of most probable expectations) discounted for uncertainty. This, how- 
ever, comes rather close to making the problem of uncertainty disappear 

It might be preferable to leave these "most probable" revenue and cost 
functions, and the marginal efficiency functions, undiscounted, and to 
take into account separately that producers do not typically aim at the 
simple objective of maximizing "most probable" profits (i.e., at equating 
the relevant marginal functions). They aim at some reasonable compro- 
mise between "most probable" profit-maximization and safety, where 
"most 'probable/' of course, merely stands for something in the nature of 
a "best guess." Safety considerations alone would usually justify a lower 

68 However, for explicit discussion of these problems cf. G. L. S. Shackle, of. cit., and 
his exchange of views with A. G. Hart in the October 1940 issue of the Review of 
Economic Studies; also A. G. Hart, Anticipations, Uncertainty and Dynamic Planning 
(Chicago, 1940). For further references cf. the present writer's op. cit., p. 152. 

09 See also J. S. Bain, pp. 154-157, below. 


rate of output 70 than that warranted by "most probable" profit-maximiza- 
tion, because the gap between "most probable" average revenue and 
"most probable" average cost (i.e., the gap between the AR and the AC 
function of value theory) typically is at a maximum for lower rates of 
output, and this is the gap— this is the "safety margin"— that counts for 
most safety considerations. Safety margins here are -meant to express the 
margins by which the actual outcome may fall short of the best guess 
without causing losses. 71 Consequently, if uncertainty increases, and 
higher safety margins are required, those producers who stay in business 
will ceteris paribus reduce their output (in the direction of higher safety 
margins), and some producers will go out of business because the safety 
margins available to them are insufficient even for the "safest" rate of out- 
put. This, of course, is not the only way in which uncertainty can be 
taken into account. In fact, it is a crude way, but it may be adequate for 
certain purposes. 

(6) The problem of interest-rate rigidities can also not be appraised 
adequately without a satisfactory analysis of uncertainty. The reduced 
emphasis on interest-rate adjustments is one of the characteristic features 
of recent theorizing on output and employment. Not much confidence is 
placed in the reduction of interest rates as a means of raising aggregate 
output, and, consequently, not much emphasis is placed on "too high" 
interest rates in the explanation of unemployment. In the Keynesian 
theory this lack of confidence is explained by the alleged infinite (or ex- 
ceedingly high) elasticity of the liquidity preference function 72 at low net 
rates of interest, as a consequence of which it is said to be impossible to 
lower net rates sufficiently. Alternative explanations have not been ex- 
plored thoroughly, and this is largely due to the circumstance that other 
explanations stress the uncertainty factor, for which no adequate allow- 
ance is made in the contemporary analytical systems. If we abstract from 
uncertainty by operating with "net" functions (discounted for risk), then 
sufficiently low net rates of interest should always produce a high rate of 

70 Assuming that the producer in question stays in business at all, in spite of the 

71 The output for which most probable average revenue minus most probable average 
cost is at a maximum gives the producer a higher safety margin than the output for 
which most probable marginal revenue equals most probable marginal cost, provided he 
feels that the unfavorable surprise (against which he is protecting himself) would 
express itself in lower demand and higher cost functions than those appearing most 
probable and 'provided that the shifts of the curves are approximately parallel shifts. 
Therefore, a compromise between the maximization of most probable profits and the 
maximization of safety margins should be expected to result in a rate of output lying 
between the output which equates most probable MR with most probable MC and the 
output which maximizes the excess of most probable AR over most probable AC. 

7 " That is, of the demand for funds intended for new hoarding. 


investment, unless the (net) marginal efficiency of capital declines to 
zero. Ineffectiveness of interest-rate policies could then be explained only 
on the assumption that net rates cannot be lowered sufficiently. 73 But this 
is no very plausible explanation. It seems far more likely that net rates 
could be lowered, practically, to any extent desired but that this, in itself, 
may fail to produce a high level of output. However, if this is maintained, 
then the emphasis must be placed on uncertainty. 74 In the first place, 
even with net rates in the neighborhood of zero, the actual terms of lend- 
ing to business would still include the full premia for borrowers' uncer- 
tainty. Secondly, even if these were largely borne by some public agency 75 
and, thus, gross rates of interest were also reduced to the neighborhood of 
the zero level, borrowing and investing would still require that the bor- 
rowers' safety margins should be sufficient in the face of the existing un- 
certainty. (It must be assumed that loan renewal depends on business 
results and is not guaranteed in advance indefinitely, because otherwise 
we would be discussing a policy of unlimited subsidization rather than a 
credit policy.) In other words, "most probable" yields (the "best guess" 
lor yields) from real investment would still have to be in excess of the 
gross rates of interest by a sufficient margin, for a sufficient number of 

The main point here is that the limitations of interest-rate policies are 
connected with a much broader area than that which is visible to an in- 
vestigator who discounts his functions for risk and then proceeds as 
though the resulting net profit expectations were maintained with cer- 
tainty. Such an investigator cannot easily explain why a sufficient reduc- 
tion of interest rates should not be the panacea for which the world is 
looking in periods of unemployment. Therefore, he will be inclined to 

73 Unless it is maintained that the marginal efficiency of capital had already declined 
to zero! 

74 In the Keynesian theory of speculative hoarding (which assumes a liquidity- 
preference floor to net rates of interest through the flattening out of the liquidity 
function), the emphasis also is placed on uncertainty of a specific kind, namely, on 
uncertainty concerning future net rates of interest. The public hoards because it believes 
that net rates are likely to rise (i.e., capital values are likely to fall) and that, therefore, 
it will pay to postpone security purchases. This is the only kind of uncertainty that is 
not defined away from the Keynesian system by "discounting for risk." All basic func- 
tions of the system are "discounted" and, therefore, they are "net of risk." Consequently, 
this specific kind of uncertainty has to bear the full burden of the Keynesian theory of 
hoarding. Our point is that if we do not get rid of other kinds of uncertainty in this 
fashion, then these other varieties— especially those attaching to profit expectations— can 
be made to share the burden. Hoarding which arises from this kind of uncertainty (pre- 
cautionary hoarding) is not actually excluded from the Keynesian analysis, but it plays 
a subordinate and passive role because the phenomenon which gives rise to it is made 
to disappear in the formal system by "discounting for risk." 

75 Which, however, implies subsidization on a truly large scale. The feasibility of such 
a scheme is questionable. 


argue that the difficulties arise merely from the impossibility of lowering 
net rates of interest sufficiently. An increasing number of economists feel 
that many crucial questions cannot even be asked in such a framework. 

(7) Conditions are favorable to investment if, for a substantial num- 
ber of firms, the existing relationship between gross 76 revenue expecta- 
tions and gross 70 cost expectations (given the existing degree of uncer- 
tainty) is such as to make it appear unlikely that they will suffer losses 
on the investment projects which are open to them. The essence of the 
optimum problem raised in connection with wage theory (pp. 79-80) 
may now be restated in a few sentences. An increase in profit margins 77 
beyond certain limits becomes self-defeating because uncertainty grows 
beyond all reasonable limits if the average propensity to consume falls to 
very low levels. Consequently, beyond a certain limit, the same factors 
which give rise to an increase in the safety margins produce an even 
greater increase in the degree of uncertainty, so that the relationship of 
these to one another becomes increasingly unfavorable. At the other end 
of the scale (i.e., for low profit margins), the opposite is true. Here, an 
increase in profit margins is favorable in spite of the lowering of the aver- 
age propensity to consume, because a high average propensity to consume 
merely reduces but does not eliminate uncertainty, and, hence, it alone 
is insufficient to call forth an adequate rate of investment if profit margins 
fall below certain limits. 

(8) In the present section, we considered so far two aspects of the 
problem of rigidities : the problem of wage rates and that of interest rates 
in relation to the requirements of full production. These problems ap- 
pear as problems of "rigidity" if they are viewed in relation to full produc- 
tion because what matters, from this point of view, is the failure or the 
inability of these cost factors to "adjust" in such a way as to clear the 
market. The rigid behavior of cost factors is, of course, connected with 
monopoly power. Monopolistic tendencies have aggravated these rigidi- 
ties considerably. However, rigidities in this sense are not exclusively a 
product of monopoly. In the preceding paragraphs it was argued that 
downward corrections of the wage level and of interest rates would not 
always create full production. Downward corrections of the wage level 
beyond certain limits may even exert an adverse influence; downward 
correction of interest rates seems always desirable in periods of unem- 
ployment, but such a measure— even if applied to gross rates— would still 
leave a limited number of investors with limited safety margins, 78 and, 

76 Gross, in the sense of "undiscounted for risk." 

77 And, thereby, in safety margins. 

' 5 Which depend on their expectations with respect to revenue and other costs. 


therefore, it may fail to result in full production. Consequently, the ab- 
sence of monopolistic rigidities would not automatically produce a cost 
behavior by which the market is always cleared. The absence of monopo- 
listic rigidities fius uncertainty would/ 9 but this is a different proposition. 
If by "rigidity" we mean failure to adjust in such a way as to clear the 
market, then rigidities are created partly by monopoly and partly by un- 
certainty, or rather by the interaction of these two elements. In this sense, 
the cost structure could show "rigidity" even if there were no monopoly. 
Uncertainty, in itself, may prevent gross interest rates from declining to 
the levels required for full production, 80 and wage adjustments beyond 
certain limits may not remedy such a situation. 81 

(9) The role of uncertainty should be stressed also in connection with 
rigidities in the commodity price structure, that is, in connection with 
rigid "relative prices." The views which have been expressed on this 
problem are partly contradictory. There seems to be good logic, as well as 
common sense, in the rather general contention that, in periods of under- 
employment, it would be beneficial to reduce the prices of specific com- 
modities—or specific groups of commodities— for which the demand is 
elastic. Special emphasis may be placed on groups of complementary pro- 
ducers' goods (e.g., building materials), the demand for which might 
show an elastic response in the event of a joint and co-ordinated price 
reduction, while the demand for the single commodities of which the 
group consists might not be elastic. On the other hand, the view was also 
expressed that inelasticity might have advantages, 82 because if the prices 
of commodities are reduced for which the demand is inelastic, 83 the con- 
sumer spends less money on the commodity in question and, therefore, is 
left with more money for other purchases (in spite of buying an increased 
physical quantity of the commodity in question). 

To the present writer this last view does not seem convincing. Let us 
assume at first that it actually is possible to increase the real output of a 
specific commodity by lowering its price. 84 In this event, the magnitude 
of the offsetting item (decreased physical demand for other goods) and, 
therefore, the net effect on real output depends on how elastic the de- 

79 In other words, Chamberlin's perfect (and not merely pure) competition would. 

80 The levels required for full production may in certain circumstances even be "nega- 
tive levels," implying, of course, outright subsidization. 

81 Underemployment equilibrium is, of course, excluded if flexible wage rates are 
assumed. But underemployment is not excluded, especially if we assume that uncertainty 
grows with falling real wage rates. 

82 Cf. Donald H. Wallace, "Industrial Markets and Public Policy: Some Major Prob- 
lems," in Public Policy, C. J. Friedrich and Edward S. Mason, eds. (Cambridge, Mass., 

83 In the sense of less than unitary elasticity. 

84 Owing to the fact that more is demanded at the lower price. 


mand for the aggregate of the other commodities is in relation to the 
demand for the specific commodity under consideration. The other com- 
modities are made more expensive in relation to the commodity in ques- 
tion. Therefore, there will be an adverse primary effect on the aggregate 
of the other markets. Disregarding at first the money income effect, we 
may conclude that the primary impact on real output will be favorable if 
the elasticity of the demand for the cheapened commodity is greater than 
the elasticity of the demand for the aggregate of the other commodities 
(i.e., greater than the average price-elasticity in the economy). In 
addition, there will be an expansionary money income effect (money ex- 
penditures will rise) if the elasticity of the demand for the cheapened 
commodity is greater than unity. Consequently, on these assumptions, a 
case can be established for the reduction of the "sticky" prices of commod- 
ities for which the demand possesses more-than-average and greater-than- 
unitary elasticity. More-than-average elasticity means that the primary 
adverse effect on the other markets is smaller than the favorable effect on 
the specific market in question; and more-than-unitary elasticity means 
that the repercussions via money income are also favorable. The question 
of whether the buyer is left with more money after completing the pur- 
chase of the cheapened commodity is not in itself decisive, because in this 
type of analysis it. may not be taken for granted that, if "left with more 
money/' the buyer will take more of the other commodities. Whether or 
not he will do so is "decided" by the assumptions we make with respect 
to the relative elasticities involved in the problem. 

As was stated before, this kind of reasoning assumes that it is possible 
to increase the output of a commodity by reducing its prices. This is not 
necessarily true. In perfect competition, it would never be true (in the 
long run). But the problem should be considered in the context of im- 
perfect competition, with a substantial area of monopolistic pricing. 
Whether monopoly output can or cannot be increased by administrative 
lowering of the price depends on the safety margin considerations to 
which reference was made in earlier parts of this section. In the first 
place, it is, of course, possible that the enterprise in question will give up 
producing the commodity (instantaneously, or in the long run), because, 
at the lower price, safety margins against unfavorable surprise are insuf- 
ficient. Secondly, the enterprise may go on producing, but at a lower rate, 
because— given the price ceiling— safety margins 85 are considered ade- 
quate for a lower output, but not for the previous output, and even less 

HiJ E.g.. in the sense of the gap between AR and AC, interpreted as "most probable" 
average revenue and "most probable" average cost (or the "best guess" for AR and AC, 
respectively; c£. p. 81, above, including footnote 71). 


for the higher output which the authority hoped to call forth. 86 Thirdly, 
intervention of this sort may result in the desired increase in output, so 
far as the processes of production are concerned which are already carried 
on in the economy, and yet it may exert an adverse influence on the plan- 
ning of new processes. This again is a matter of uncertainty. Only in cases 
in which these three qualifications may be ruled out as practically unim- 
portant, should we expect a favorable "aggregate output effect" from the 
lowering of specific prices. In these cases, the elasticity considerations of 
the preceding paragraphs become relevant. 

In conclusion, it should be repeated that the process of including prob- 
lems of the cost-price structure in the theory of employment is still in its 
early stages. The theory of employment is frequently made to proceed on 
the assumption of a given cost-price structure, and it is then concerned 
almost exclusively with aggregate money flows, without investigating the 
relationship between these, on the one hand, and cost-price problems, on 
the other. At the same time, value theory typically proceeds on the as- 
sumption of a given aggregate output and employment, and it is con- 
cerned merely with the relative allocation of resources. It is to be expected 
that the links between the two will grow tighter and that significant inter- 
actions will be explored more fully. 

IV. Policies 

(i) In the recent discussion of full-employment policy, the emphasis 
has been placed increasingly on compensatory fiscal devices. 87 This ac- 
cords well with the present state of the theory of employment, as inter- 
preted in the preceding sections of this chapter. 

Compensatory fiscal policies are, of course, not "cost-price policies"; 

86 As my colleague, Professor Joe S. Bain, pointed out to me in discussion, these are 
cases in which the price reduction leads to rationing by the producer. The demand is 
always higher at the lower price. The existence of excess demand, of course, reduces the 
uncertainty but the safety margins must be sufficient to compensate for the remaining 
uncertainty as concerns shifts of the relevant functions. 

87 That is to say, on fiscal policies aimed at stimulating private expenditures and at 
supplementing them by public expenditures when increased aggregate expenditures seem 
desirable; and at reducing private as well as public expenditures in inflationary periods. 

Full employment may be defined as a condition in which no person is unemployed 
who desires employment at the going money wage rates. In practice the closest conceiv- 
able approximation to full employment is a condition in which employment is "full," 
aside from a moderate amount of "frictional unemployment." Some degree of arbitrari- 
ness is involved in the concept of frictional unemployment because in many cases there 
exists no sharp distinction between the inability and the unwillingness to "move" 
(regionally or occupationally), and unwillingness to move cannot always be sharply 
distinguished from unwillingness to work at the going wage rates (i.e., from voluntary 
unemployment). There may exist any amount of voluntary unemployment in "full 


they do not aim at influencing the cost-price structure, although, indi- 
rectly, they may influence it. The theory of the relationship between 
cost-price problems and the level of employment is comparatively unde- 
veloped and so is the corresponding policy discussion. The opposition to 
neglecting this aspect ot the full-employment problem has gradually 
grown (cf. the work of Clark, Ellis, Haberler, and others), 88 but the main 
trend of thought has so far mostly proceeded along ' monetary-fiscar 
lines. This is not entirely accidental, nor does it merely reflect the un- 
willingness to invoke the resistance of power groups against interference 
with the cost-price structure. The problem of the effects of cost-price 
policies is inherently more complicated than that of the effects of tax in- 
creases and reductions, or of changing rates of public investment. 

(2) The most obvious (although not the only) difficulty which arises 
if cost-price problems are disregarded in connection with compensatory 
fiscal policies relates to the inflation problem. To "compensate" deflation 
tendencies, with uncompromising consistency, means to guarantee the 
continued existence of a sellers' market without interruptions. In such 
circumstances, a continuous upward pressure on the general wage and 
price level would have to be expected. Furthermore, there would presum- 
ably develop a tendency toward quality deterioration, and pressure would 
be exerted to adjust the fiscal program to any existing maldistribution of 
resources. Inflationary tendencies would be generated by the concerted 
wage- and price-raising action of organized groups, because the economic 
penalty standing in the way of such action would be removed. 89 These 
tendencies would be reinforced by the dishoarding of idle balances (the 
maintenance of which would appear to be unjustified under such condi- 
tions). It is true that a group of producers which raises its selling prices 
would place itself at a competitive disadvantage in relation to other 
groups as long as these do not follow suit. But this would not check the 
inflationary tendency because— given the de facto guarantee of full em- 
ployment—labor groups would always prefer higher money wage rates to 
the status quo, and, consequently, producers would usually be better off 
with higher prices, regardless of whether the other industries actually fol- 
lowed suit. Furthermore, it could be taken for granted that other indus- 
tries, for the very same reasons, would follow suit. 

It has been maintained that a national bargaining agency for the ag- 
gregate labor force would not behave in the manner here assumed be- 


88 Cf., e.g., J. M. Clark, "Financing High Level Employment," in Financing At 
can Prosperity: A Symposium of Economists (New York, 1945), pp. 71-125; Howard 
S. Ellis, "Economic Expansion Through Competitive Markets," ibid., pp. 126-198; 
Gottfried Haberler, op. cit. 

80 Cf. A. C. Pigou, op. cit. 


cause the wage increases envisaged above are obviously self-defeating for 
labor as a whole. 90 But this is more than questionable. Even an over-all 
bargaining agency for labor would consist of constituent groups, and the 
agency would usually find it more expedient to yield to each group in 
succession than to oppose all. Aside from this, the agency would always 
tend to maintain that the wage increases it is demanding should not 
result in price increases, while in reality they would. 

(3) These are the main difficulties arising in connection with a de 
facto guarantee of full employment by means of fiscal policy. The range 
of attitudes that may be taken in an attempt to resolve these difficulties 
lies between two extremes. One extreme position would be that of advo- 
cating thoroughgoing and rigorous controls, extending to prices, wage 
rates, the quality of goods and services, and the regional and occupational 
composition of resources (including the labor force). Total economic 
controls could be handled effectively only by a government which itself 
is the only effective power group in the scene. The price is excessive. 

At the other extreme, we have the laisser faire attitude in relation to 
cost-price problems. It is important to realize that a laisser faire attitude in 
this respect is compatible with a substantial amount of "compensatory 
fiscal policy." For example, it would be quite unconvincing to argue that 
during the Great Depression unregulated cost-price relationships pre- 
vented the government from adopting vigorous compensatory policies of 
a monetary-fiscal character. On the contrary, failure to adopt policies of 
this kind on a sufficient scale contributed greatly to the growth of cost- 
price maladjustments. But while cost-price laisser faire does not justify 
monetary-fiscal laisser faire, it does set limits to the effectiveness of com- 
pensatory fiscal policy. With unregulated cost-price relationships, fiscal 
policy must keep one eye on the full-employment problem and the other 
on the wage and price level. It cannot keep both eyes on the full-employ- 
ment problem. Expansionary monetary-fiscal policies must be discon- 
tinued if they result in substantial cost-price maladjustments of the 
"inflationary" variety, and this may happen at levels considerably lower 
than that of full employment. 

(4) Even though the United States does not at present have com- 
plete cost-price laisser faire, we are not far removed from this condition. 
Some existing measures (even aside from the temporary rent control) 
may be regarded as falling in the category of wage and price controls, but 
these are not major features of the present institutional setting. What 
degree of cost-price control will prove to be desirable and compatible with 
democratic political institutions is hard to foretell. As was pointed out 

' M Sir William Beveridge, oy. cit. 


before, this problem deserves more attention than it has received. How- 
ever, it seems very likely that there exists a fundamental contradiction 
between truly thoroughgoing controls of this kind and democratic politi- 
cal institutions, although it also is very likely that more could be done to 
reduce the exploitation of man-made (or "institutional") scarcities. With 
a largely uncontrolled cost-price structure, monetary-fiscal policy will 
have to be oriented to "full-employment" objectives, on the one hand, 
and to "wage and price stability" objectives, on the other. It cannot be 
oriented to full-employment objectives alone. Such a policy can be suc- 
cessful only if moderate concessions at the expense of one set of objectives 
result in coming reasonably close to accomplishing the other. The prob- 
lem is inherently one of balancing objectives. This is overlooked in much 
of the recent full-employment literature. 

Despite these limitations, compensatory fiscal policy can go a long way 
toward eliminating periods of mass unemployment. It is to be hoped that 
the limitations so far discussed will mainly express themselves in an occa- 
sional dilemma at tolerably high levels of employment, where it will still 
be necessary, at times, to choose between some amount of cyclical unem- 
ployment and highly undesirable wage-price tendencies. It would be un- 
warranted to anticipate such a dilemma for thoroughly unsatisfactory 
levels of aggregate output and employment. Compensatory fiscal policy 
should be capable of going a long way toward preventing periods of mass 
unemployment, even if it is incapable of preventing business recessions. 
This, of course, would be a tremendous accomplishment. Most major de- 
pressions might not have developed beyond the stage of "tolerable" reces- 
sions if effective compensatory measures had been adopted six to twelve 
months after the downturn, and extended periods of stagnation (or of 
chronic depression) may be interpreted as aftermaths of violent and un- 
compensated cyclical contractions. 

(5) It is conceivable that this high promise of monetary-fiscal policy 
induces many economists to underemphasize, or even to disregard, the 
limitations previously discussed. The fear is rather common that the pub- 
lic and the authorities will not realize the possibilities offered by the 
appropriate devices. Some degree of optimism may, however, not be un- 
justified in this respect. Looking at the other side of the medal, we see 
that compensatory— in this case anti-inflationary— fiscal policies were car- 
ried distinctly further during the Second World War than in previous 
major wars. These fiscal policies were far from adequate, but they were 
Jess inadequate than might have been expected on the basis of the experi- 
ence of previous inflationary periods. On the whole, the history of the 
Second World War shows an increasing (but still inadequate) under- 


standing of the issues involved in compensatory fiscal policy. Moreover, 
it should be taken into account that strategic power groups frequently 
have reason to believe that the inflationary distribution of a war burden 
will prove more favorable from their point of view than a strictly planned 
distribution. A major depression, however, is a disaster for all groups. 

(6) What, in particular, are the fiscal devices most frequently ad- 
vocated in recent literature? Flexible tax rates and the countercyclical 
timing of ' noncompeting ' public works 91 have received much attention, 
and rightly so. Insofar as the public works required for overcoming de- 
pression tendencies are useful fer se— i.e., do not merely serve the pur- 
pose of generating a Multiplier Effect— the choice between reducing tax 
rates and stepping up the public works program becomes a matter of 
"opportunity costs," in a somewhat extended sense of the term. If, on the 
other hand, the public works in question are highly wasteful in any sense 
other than that of generating a Multiplier Effect, it seems preferable to 
accomplish the desired objective wholly by reduced taxation, and pos- 
sibly by granting consumer subsidies (that is to say, by negative taxa- 
tion). The urgency of certain "noncompeting" (or largely noncompet- 
ing) public works projects 92 is obvious at present. One might believe that 
for the next one or two decades the alternative method (reduced taxa- 
tion) should be disregarded. However, the most urgent public works are 
not those which are most suitable for prompt adjustment to rapidly 
changing business conditions. Moreover, the response of the economy to 
the stimulus provided by public works is substantially reduced by high 
taxes. Consequently, it seems highly desirable to combine the two meth- 
ods. It is recognized in the contemporary literature that further advance 
in the field of social security, and especially of unemployment insurance, 
would have the desirable by-product of lending more flexibility to the tax 
structure, because contributions are in the nature of taxes, and benefits in 
the nature of "consumer subsidies." But it also is necessary to lend tax 
rates substantially more flexibility. 

The adjustment of tax rates to changing business conditions would 
presumably require (in addition to the pay-as-you-go system) some dele- 
gation of power on the part of Congress, because tax revision has proved 
a very time-consuming procedure. This also is recognized in the contem- 
porary literature on the subject. The proposal has been made that Con- 
gress should delegate to the Administration the power of changing the 

531 That is, of public works which do not suppress a comparable amount of private 
investment, or, if possible, even stimulate private investment activity directly. 

92 Especially that of medical and educational projects, of slum clearance and low-cost 
housing, of projects aiming at the conservation of resources, road building, etc. 


first-bracket individual income tax rate during the fiscal year. 93 Possibly, 
it might even be sufficient to have Congress delegate this power to a joint 
committee of its own. 

(7) The fiscal policy proposals so far considered may be viewed as 
compensatory devices. This is not their only raison d'etre, but, in addition 
to having other merits, they would perform the function of providing 
stimuli in periods of recession, and of checking inflationary tendencies. 
Other fiscal policy proposals widely discussed in the contemporary litera- 
ture aim at strengthening the inducement to invest, aside from the prob- 
lem of cyclical instability. The opinion is widely held that corporate taxes 
are less desirable instruments of fiscal policy than individual income 
taxes, not merely because individual income taxes are more equitable, 
but also because business management, in deciding about investment 
projects, is much more concerned with profits "after corporate taxes" 
than with dividends "after individual taxes." 94 However, if individual 
income taxes were to be substituted for the present type of corporate 
taxation, it would be necessary to find means by which undistributed 
profits become accessible to taxation. Otherwise, an unreasonable stimu- 
lus would be provided for accumulating undistributed profits. This 
would be adverse to the interests of stockholders, and it would have 
especially bad general economic effects in periods of business reces- 

Yet the deterrent effect of individual taxes on the willingness of in- 
dividuals to invest should also not be overlooked. This adverse effect— 
and, generally speaking, the deterrent effect of taxation— could be re- 
duced by appropriate loss-deduction (or loss carry-over) provisions. The 
problem of "averaging devices" of this sort (i.e., of devices aiming at the 
averaging of income over longer periods, for tax purposes) has also re- 
ceived considerable attention in the recent literature. 

(8) The impression should not be conveyed that the idea of compen- 
satory fiscal policy is wholly the product of the last ten or fifteen years. It 
is more appropriate to say that the idea of compensatory credit policies 
(or "monetary" policies in the narrower sense), as well as that of compen- 
satory fiscal policy, has a long history, but that recently the emphasis 
has been shifted increasingly from credit policies to fiscal policy. The 
limitations of credit policies— such as those of discount policy and of 
open-market policy— are generally assumed to be narrower than those 
of fiscal policy. The main reason for this is that additional central bank 

88 Cf . Committee for Economic Development, Jobs and Markets (New York and Lon- 
don, 1946). 

94 Cf . Harold M. Groves, Production, Jobs and Taxes (New York and London, 1944). 


credit may, in certain circumstances, result in a comparatively large 
increase in hoarding and only in a small increase in investment, while 
it seems safe to assume that the additions to the income flow (rather than 
merely to the money stock) which are generated by expansionary fiscal 
policy 95 give rise at least to the "corresponding" increase in consumption 
expenditures, and probably also to induced investment. 

Furthermore, if inflation rather than depression tendencies need be 
compensated, tax increases and budgetary savings can (jointly) always 
be made sufficient, while any increase in interest rates (plus credit 
restriction) may prove insufficient, provided the public owns enough 
idle balances. Also, governments can ill afford to bring about really 
significant increases in interest rates during inflationary periods, because 
the attending capital losses of bondholders would presumably make 
borrowing difficult in later periods in which this might again become 
desirable or necessary. Hence the shift in emphasis from "monetary" 
policy in the narrower sense to fiscal policy. 

(9) There is some correspondence between this shift in emphasis, 
on the one hand, and the shift from the quantity theory approach to the 
savings-investment approach, on the other. 96 In the quantity- theory 
terminology, both the credit policies and the fiscal policies result in the 
creation or destruction of money stocks, and, consequently, the difference 
between the two types of policy is not emphasized "from the outset." In 
the savings-investment (or income-expenditure) terminology, fiscal pol- 
icy changes the "strategic" variables (such as income itself) instantane- 
ously, while credit policies do not directly or necessarily affect these 
magnitudes. However, any assumption concerning facts can be ex- 
pressed in either conceptual framework. There obviously is no logical 
inconsistency in maintaining— in the framework of the quantity-theory 
approach— that the velocity of new money is different, depending on 
whether it enters through open market operations or through deficit 

Earlier in this chapter, the view was expressed that the shift in 
emphasis from the quantity theory approach to the savings-investment 
approach has proved fruitful, but that this shift may have been some- 
what too complete. Substantial changes in the stock of money may well 
have an influence on the basic relationships (functions) of the savings- 
investment approach, and liquidity ratios need not be in the nature 
of mere residuals, even at low interest rates. A similar statement could 
be made of the shift in emphasis from "monetary" policies in the nar- 

95 E.g., the additional disposable income created by tax reductions or by public works. 
w Cf. pp. 52-53, above. 


rower sense 97 to fiscal policy. This shift in emphasis has also proved 
fruitful: fiscal policy is the more potent instrument, and the reasons, 
by and large, are those which were brought out in the literature of the 
last decade. But this shift in emphasis also tends to become unduly 
complete. It leads to overlooking the fact that in the past even the mone- 
tary policies (in the narrower sense) were not carried far enough and 
that nobody can tell what additional effects might have been obtained 
by carrying them further. The facts do not really prove that these 
policies are inherently inadequate, regardless of the scale on which they 
are undertaken. For example, during the Great Depression, open-market 
purchases were undertaken on an entirely insufficient scale. The stock 
of money was allowed to shrink violently, especially from the early 
part of 1 93 1 on, which is all the more noteworthy because, at that time, 
signs of an incipient recovery were observable. Not only was the fiscal 
policy of the Great Depression inadequate, but the monetary policy 
was also. The same is true of inflationary periods. 

(10) Opponents of the "modern tendencies" in this field usually 
emphasize the complications to which a large public debt may give rise, 
the "socialist" implications of public works projects, and the "superficial" 
character of monetary-fiscal remedies. None of these objections are 
completely unfounded, but it is likely that the appropriate compensatory 
policy could reduce the validity of these arguments to such an extent 
that the advantages of the policy would be entirely out of proportion 
to the valid elements of these objections. 

( 1 1 ) As for the public debt problem, this is a matter of internal trans- 
fers from taxpayers to the owners of government securities. In a full- 
production economy, this transfer burden is increasing, relatively to 
income, only in the event that the public debt rises more rapidly than 
productivity, and, thereby, more rapidly than aggregate real output. 98 
On realistic assumptions, this seems unlikely. If, in the long run, periods 
with expansionary tendencies and periods with deflationary tendencies 
stand in a one-to-one ratio to each other (so far as average duration and 
vigor are concerned), compensatory fiscal policy does not lead to any 
long-run increase of the public debt. Such a one-to-one relationship 
must, of course, not be taken for granted. However, even a substantial 

67 I.e., from credit policies or interest-rate policies. 

98 Assuming a constant interest rate and a constant price level. Cf. Evsey D. Domar, 
"The 'Burden' of the Public Debt and National Income," American Economic Review, 
December 1944, XXXIV, pp. 798-827. When maintaining that a more rapid rise of the 
debt than of the tax base is highly improbable, we think of a rise in the debt such as is 
produced by compensatory fiscal policy. A national emergency would of course be a dif- 
ferent matter (cf. footnote 10 1). 


long-run excess of deflationary over expansionary tendencies may prove 
compatible with a rate of increase in the public debt" which falls dis- 
tinctly short of the probable long-run increase in aggregate productivity 
and, eo ifso, of the long-run rise in the yield of given tax rates. 100 Further- 
more, if the rise in the internal transfer burden, which is caused by the 
rise of the public debt in deflationary periods, should nevertheless 
threaten to grow to an alarming size, 101 it always should be possible to 
borrow directly or indirectly from the central bank. 102 This actually 
would change the internal transfer in question to a mere bookkeeping 
item, in addition to possessing other advantages. 103 Such operations 
become a variety of interest-free borrowing, provided the profits of the 
central bank, over and above a stated rate, are conducted back to the 
Treasury. The disadvantage of this solution may be that it forces the 
commercial banks to handle additional deposits without giving them 
additional earnings. 104 But there surely should be some way of over- 
coming this difficulty. To say that there is none, means to maintain 
that an economy is incapable of getting out of a depression because it 
is impossible to decide who should bear the costs of handling the addi- 
tional deposits which must be created to offset the deflationary tendency! 
(12) The opposition to compensatory fiscal policy rests only in part 
on the complications which might arise in connection with the debt. 
It is sometimes maintained— or at least the impression is conveyed— that 
large-scale public works projects do not accord with the principles of 
free enterprise. Yet in all countries (including the United States), 

59 Such as is produced by compensatory fiscal policy. 

100 Eo ipso, due to the progressive character of the federal tax structure. 

101 In a national emergency the debt would be rising much more rapidly than the tax 
base if the methods of financing should be such as in the past. An emergency would 
call for very severe taxation, and— to the extent to which inflationary borrowing is politi- 
cally unavoidable— for interest-free borrowing from the central banks or for very cheap 
borrowing from the commercial banks. On realistic assumptions, compensatory fiscal 
policy (in contrast to a national emergency) is almost certain to be compatible with a 
gradual reduction of the debt burden in relation to national income and the tax base. 
This of course is partly a consequence of the high "initial" burden with which we start 
the postwar period. 

102 Indirect borrowing means, in this case, the coupling of government borrowing from 
the market with equal security purchases of the central bank on the market. 

103 Borrowing from the public or from the commercial banks need not result in the 
borrowing of idle money for government expenditures (as is intended in deflationary 
periods). Instead, it may result in the borrowing of money which would have been used 
by the public or by the bank for other purposes. The answer depends on the interest- 
elasticity of liquidity provisions. Direct or indirect borrowing from the central bank is not 
subject to this limitation. 

104 Because on these assumptions the commercial banks do not own the newly issued 
government securities. It is true that they own excess reserves, but it is implied that this, 
of itself, does not result in credit expansion. If it does, then we are faced with a milder 
deflationary tendency which can be handled by very much simpler methods, namely, by 
open market purchases or by the lowering of reserve requirements. 


there exists at present a significant backlog of urgent public works which 
have nothing to do with "socialism," on any reasonable definition of 
this term. If, at some future date, we should run out of these projects, 
and the next items on the list should be either wasteful (except for the 
Multiplier Effect) or if they should suppress a comparable amount of 
private investment, then tax reductions become clearly superior means 
of combating depression tendencies. In such hypothetical conditions, if 
far-reaching tax reductions should not generate a sufficient expansionary 
effect, then consumer subsidies would have to be adopted. With the 
exception of adequate unemployment compensation, such subsidies might 
actually give rise to difficulties of a more fundamental character, because 
of the arbitrariness involved in distributing the funds, and because of 
the adverse influence which free gifts might exert on the supply of 
services. But at present these are remote worries, and, in all probability, 
they will remain remote for a long time to come. 

(13) Compensatory fiscal policy does have socialist implications— and 
it probably is incompatible with the present type of political institutions— 
if it takes the form of a full-employment guarantee and if a rigorous and 
comprehensive system of direct controls is adopted to prevent the infla- 
tionary consequences which would otherwise result. But the policy can- 
not justly be said to have these implications if it is geared to aggregate 
employment objectives, on the one hand, and to wage- and price-level 
objectives Qprice stability objectives"), on the other. In this event, compre- 
hensive direct controls would not be required to forestall inflation and 
to prevent extreme maladjustments in the cost-price structure. This is 
a very substantial advantage, even though it is acquired at a cost. 105 The 
cost expresses itself in the fact that the policy can be successful only to 
the extent to which the high-employment objective is complementary 
to, or at least compatible with, the objective of a reasonably stable and 
reasonably free price level. But while, in these circumstances, compen- 
satory policies are indeed subject to this limitation, they would probably 
prevent such developments as occurred during the last two or three years 
of the Great Depression and its aftermath. 

(14) The objection that monetary and fiscal policy is "superficial" 
and that it does not penetrate to the roots of the maladjustments is some- 
what vague but essentially correct. An approach which disregards the 

105 Cf. David McCord Wright, The Economics of 'Disturbance (New York and Lon- 
don, 1946). Professor Wright draws a contrast between social-economic systems charac- 
terized by rapidly changing methods of production, by economic fluctuations and by 
comparatively easy access to political power, and systems characterized by economic 
stability (including a tendency toward the "stationary") and by self-perpetuating power 
groups in the political scene. 


interaction between cost-price problems, on the one hand, and aggre- 
gate output problems, on the other, is seriously incomplete on the theo- 
retical level, and it leads to compromise solutions on the level of policy. 
The "cost-price" problems are mainly those arising from institutional 
scarcities (producers' monopoly power and unionism), from uncertainty, 
and from immobility, all of which mutually intensify each other. It 
would be highly desirable to explore more fully these interactions and 
their bearing on the problem of aggregate output and employment, and 
to examine more fully the types of policy that might have a favorable 
effect on cost-price behavior. For, while the Office of Price Administra- 
tion and the War Labor Board would be unacceptable as peacetime 
agencies— and while, even if they were "acceptable," they would be run 
most ineffectively by a democratic government— it does not follow that 
nothing can be undertaken to prevent the full exploitation of institutional 
scarcities. All this points to areas of research which so far have remained 
comparatively undeveloped. But it does not justify the position that we 
should refrain from using the available weapons merely because our 
armory is incomplete. 

The GNP Model: An Illustration 

(1) It is usually possible to discern the progress achieved and also 
the difficulties not yet overcome at different periods in the development 
of fields of inquiry by examining the technical tools which are used con- 
temporaneously. So far as the theory of employment is concerned, the 
GNP (Gross National Product) model provides an illustration. This 
model was developed during the period discussed in this volume. The 
technique suggested by the GNP models is as follows. The GNP may 
be viewed as the sum total of gross incomes, or, alternatively, as the 
aggregate gross 106 output of an accounting period. Let us first view it 
as aggregate gross income. From this gross income it is possible to derive 
(net) national income by certain adjustments, mainly by deducting 
depreciation and indirect business taxes. From national income, by 
further adjustments, we derive personal income, i.e., aggregate indi- 
vidual income payments. 107 If from these we deduct personal taxes, we 

106 "Gross" in the sense that depreciation of plant and equipment is not deducted. But 
the materials and services used up in the production of other items of current output are 

107 On the way from national income to individual income payments (personal in- 
come), the main items of deduction are corporate profit taxes, net corporate savings 
(i.e., undistributed profits), and social security contributions, while the main addition 
is for transfer payments (which are treated as "income" by the recipient but which do 
not constitute a compensation for a contribution to current output, such as veterans' 


arrive at the disposable income of individuals. Part of the disposable 
income is spent on consumption, while the remainder is saved by indi- 
viduals. By deducting the savings of individuals from the disposable 
income, we arrive at the value of one of the items of which the GNP con- 
sists when viewed as output rather than income, namely, at the value 
of the current output for ultimate consumption. 108 So from now on, we 
view the GNP as aggregate output. It consists of the "ultimate" con- 
sumption output just derived, plus private gross capital formation, 1 " 9 
plus the government expenditures on goods and services produced dur- 
ing the period. Consequently, if we start from the GNP viewed as gross 
income, and move down, by the appropriate deductions, via disposable 
income to consumption expenditure, and then add private gross capital 
formation and government expenditures on currently produced goods 
and services, then we must obtain once more the GNP figure from which 
we started. The deductions (savings plus taxes) must always equal the 
additions (private investment plus government expenditures). The mag- 
nitudes included in the model are so defined that this instantaneous 
relationship must hold true. However, what does it mean to interpret 
the level of output of a period and its determinants, in these terms? 

(2) Such an interpretation of aggregate output and its determinants is 
an application of the ex post variety of the savings-investment ap- 
proach. 110 If no further assumptions are introduced, all valid statements 
which can be made in this framework are mere truisms, and it would be 
unreasonable to expect any degree of stability from the relationships ex- 
isting between the magnitudes included in the model. Significance at- 
taches to these relationships only if specific assumptions are introduced. 
For example, if we assume that income expectations are correct, or that 
errors tend to cancel out, then the relationship between disposable income 
and simultaneous consumption expresses a deliberate attitude, a true pro- 
pensity (not merely an ex post ratio which is defined as a propensity). 
Relationships of this kind might show some degree of stability. Similarly, 

benefits, unemployment benefits, etc.) and for interest payments by the government 
which are also interpreted as transfer payments but are distinguished from the others. 

108 If more than the current output is consumed, then this expresses itself in negative 
capital formation. Consequently, in computing aggregate output, we may treat the magni- 
tude derived in the text as an item of current output. Aggregate output includes capital 
formation which may contain negative items. 

109 Consisting of construction (i- e -> "plant" or buildings), producers' durable goods 
(i.e., equipment), the net increment in business inventories, and the net increment in 
claims against foreign countries. The concept is called "gross" because the depreciation 
of plant and equipment is not deducted. 

110 Cf. A. G. Hart, " 'Model Building' and Fiscal Policy," American Economic Review, 
September 1946, XXXV, pp. 531-558; Jacob Mosak, "National Budgets, and National 
Policy," ibid., March 1946, XXXVI, pp. 20-43; a ^° rejoinder and final reply, ibid., 
September 1946, pp. 632-640. 


if we assume that there is no unintentional accumulation or reduction of 
inventories, then capital formation mav also be interpreted as expressing 
the results of investment plans. For the short run— and, therefore, with 
respect to the problem of business fluctuations— these assumptions would 
surely be unrealistic. With respect to long-run average relationships, they 
may be less unrealistic, but the long-run expectations themselves are sig- 
nificantly influenced by "short-run" disturbances, and so is the "secular 
trend." Also, the framework itself does not directly suggest relationships 
between cost-price behavior, or the supply of money, on the one hand, 
and the explicit variables of the GNP model, on the other. 

Theorizing in terms of such models well illustrates the type of analysis 
applied in much of the contemporary literature in the theory of employ- 
ment. We have, on the one hand, quantitative frameworks ("rigorous" 
within their own limits) which can be filled with empirical data. The 
GNP model is an example of such a framework. On the other hand, we 
have the somewhat abstract description of simplified model-processes 111 
by which the "rigorous" systems must be supplemented to become suit- 
able for the analysis of significant problems. As empirical data accumu- 
late and analytical techniques are refined, some systematic framework 
may become applicable to phases of economic development which previ- 
ously could only be characterized vaguely by "purely theoretical" model 
processes. Some of the contemporary econometric models go very much 
further in this direction than the comparatively simple GNP models just 
discussed. They include psychological, institutional, and technological 
tags, by which these analytical systems are made dynamic, i.e., monetary 
equilibrium implications are avoided. Lagged relationships may prove to 
be significant and reasonably stable, even if the future is not foreseen cor- 
rectly. These complex models also include a great many more variables 
than those explicitly entering into the usual GNP models. However, no 
available formal model can claim to be considered a dependable instru- 
ment of projecting (forward and backward) and consequently we lack 
dependable criteria of superiority and inferiority. No available model is 
useful without being qualified and supplemented by less formal and often 
somewhat vague considerations and by "judgment." The degree of com- 
plexity of the models applied partly determines the nature of the appro- 
priate supplementary considerations. The question of the optimum 
combination of these elements stays a matter of individual preferences. 
So far the results obtained with the most complex models have not been 
the best. But it certainly would be a mistake to turn to the opposite 

111 Such as the period analysis discussed in earlier sections. 


J. K. Galbraith 

Excluding only the issues associated with fiscal policy and the level 
and stability of employment, no problem attracted more attention from 
economists during the 'thirties than that of monopoly. As usual, the 
sources of this interest are traceable in part to ideas and in part to circum- 
stance. Several years before the depression started, certain long-held and 
vital assumptions concerning the structure of the typical market were 
undergoing re-examination. The results became apparent at a time when 
economists everywhere were looking for an interpretation of the current 
crisis in capitalist society. A retrospect on monopoly, in its theoretical and 
applied aspects, properly begins, therefore, with a review of the ideas 
which accounted for the original revival of interest. It is appropriately 
followed by a consideration of the effect of these ideas as they were car- 
ried into the world of policy and politics. 


The first influential new step in the field of ideas was the publication 
in 1926 by Piero Sraffa of his now famous article, "The Laws of Returns 
under Competitive Conditions." 1 Subject to qualifications, the tendency 
at the time Mr. Sraffa's article appeared was to recognize the limiting 
case of monopoly, but to assume, in general, a rule of competition. Com- 
petition was not assumed to be perfect. Those already in the business 
might variously obstruct the entry of newcomers. Or entry might be ren- 
dered difficult by the prestige associated with trademarks and trade 
names. Imperfect knowledge of opportunities might interfere. Decreasing 
costs were deemed an especially serious handicap for the newcomer, 
who, because he was new, was likely to be small. If large scale and ac- 
companying requirements in capital and organization brought substantial 
economies, the small newcomer was faced with an organic handicap. 
Nevertheless, these barriers, though widely recognized, were convention- 

1 Economic Journal, December 1926, XXXVI, pp. 535-550. 



ally assumed to be of secondary effect. They were frictions that muddied 
and at times diverted but did not check the great underlying current 
which was toward a competitive equilibrium. Given that equilibrium, 
there was a presumption, again subject to many dissenting voices, that 
economic resources would be employed with maximum efficiency and 
the product so distributed as to maximize satisfactions. Sraffa attacked the 
assumption that the "frictions" were in fact a secondary and fugitive phe- 
nomenon. He argued they were stable and indeed cumulative and yielded 
a solution consistent not with a competitive, but a monopolistic equilib- 
rium. He argued that monopoly, not free competition, was the more ap- 
propriate assumption in market theory. 

In 1932-33 Mrs. Robinson and Professor Chamberlin produced the 
two books that were to become the texts for the revived interest in mo- 
nopoly. 2 The first leaned heavily on Sraffa; the second had a more inde- 
pendent genesis. Both had a prompt and enthusiastic reception. This is 
not difficult to explain. For years there had been marked discontent with 
the accustomed assumptions and the standard analysis of competitive and 
monopolized markets. Discussion and teaching had too long centered on 
what, too obviously, were limiting cases. Even (or perhaps especially) 
students were reluctant to accept the results as descriptive of the real 
world. The new work had the great advantage, from the viewpoint of 
marketability, of adding something new to something old, and of adding 
almost precisely what the customers wanted. Both books were solidly in 
the tradition of Marshallian partial equilibrium analysis; and in the 
United States and the United Kingdom this had become not only an 
utterly respectable but an all but impregnable tradition in economic 
thought. The inhabitants of this citadel, although never too hospitable to 
strangers, were bound to accept old inhabitants armed with familiar 
weapons even though they used these weapons in a seemingly dangerous 
way. Both Professor Chamberlin and Mrs. Robinson offered an organ- 
ized, intellectually palatable approach to the middle ground between mo- 
nopoly and competition— an obvious antidote to the existing uneasiness. 
In this respect their contribution was sharply distinguished from that of 
Marshall, J. M. Clark, and others who had delved but briefly into this 
intermediate area, from that of teachers who had warned their students 
that the old categories were inadequate, and from that of Cournot, Edge- 
worth, and Bowley who, at most, had offered a smattering of mostly im- 
probable solutions to mostly improbable situations. 

In retrospect the most important contribution of Professor Chamberlin 

2 Joan Robinson, Economics of Imperfect Competition (London, 1933); Edward H. 
Chamberlin, The Theory of Monopolistic Competition (Cambridge, Mass., 1932). 


and Mrs. Robinson was to emancipate the analysis of markets from the 
inadequate categories of competition (impaired by sundry frictions) and 
single-firm monopoly. Almost at once duopoly, oligopoly, and the pur- 
poseful differentiation of products became accredited and very useful 
categories in market analysis. 

This liberalization of market categories was more important than the 
theory that explored them. Professor Chamberlin did make a notable re- 
finement in the existing concept of competition. Where previously com- 
petition had often been loosely identified by the terms of rivalry in the 
market— conditions of entry, the energy and knowledge of participants, 
and the like— he in effect derived its character from the competitive 
equilibrium it was assumed to bring about. The concept of "pure" com- 
petition was thereby confined to markets where the demand for the 
product of the individual seller was infinitely elastic at the ruling price. 
This was a good deal more rigorous than existing definitions; it had, as I 
shall argue presently, important practical consequences. 

Both Professor Chamberlin and Mrs. Robinson also made the marginal 
revenue curve a standard tool of market analysis and Professor Chamber- 
lin's theory of monopolistic competition— of competition between numer- 
ous sellers differentiated by location, personality, or physical or psychic 
differences in their product— brought the vast phenomenon of merchan- 
dising and advertising within the scope of theoretical analysis. For the 
purposes of the present essay, however, it was the area of failure rather 
than achievement of the new work that is of prime significance. Without 
much doubt the dominant market of modern capitalism is not one made 
up of many sellers offering either uniform or differentiated products. 
Rather it is a market of few sellers, i.e., oligopoly. Apart from consumers' 
goods, the counterpart of few buyers associated with many or few sellers 
is also a common phenomenon. Where sellers are few the product is auto- 
matically identified with its vendor and hence there is always a measure 
of differentiation— the elasticity of substitution between products of a 
few sellers can never be quite perfect. But the ruling characteristic is the 
fewness of the sellers. 

In dealing with small numbers or oligopoly, Professor Chamberlin, 
who went farthest with the problem on a general theoretical level, did 
little more than resurrect the engaging but largely irrelevant novelties of 
Cournot and Edgeworth. This was a failure of prime importance— one 
that economists were, on the whole, slow to recognize. The failure w 7 as 
inevitable. Success, by familiar standards, implied a determinate solution. 
One certain fact about oligopoly (and its counterpart on the buyer's side 
of the market) is that the entire market solution can be altered unilater- 


ally by any single participant. This is at once the simplest and the most 
critical distinction between oligopoly and pure competition. It also means 
that the methodological device by which the competitive market has 
been analyzed, i.e., laying down general assumptions about the group 
response of numerous individuals to common stimuli, is inadmissible. 
Rather the assumptions must be sufficiently comprehensive to cover the 
behavior pattern of each participant in the market. Even though it is 
assumed that each participant seeks to maximize his return, the possible 
individual behavior patterns and resulting market solutions are almost 
infinitely numerous, and the assumption that all individuals will seek 
maximum pecuniary return (as distinct from non-pecuniary prestige, ex- 
pression of individuality, etc.) is questionable. Edgeworth and Cournot 
and, in that tradition, Chamberlin, merely derived the market solution 
that followed from two or three out of a near infinity of possible behavior 
combinations. It follows that they were not offering a theory of duopoly 
or oligopoly but displaying a few samples. Little progress has been made 
to an analysis of oligopoly by this route and little could be expected. 

The importance of oligopoly in the world as it exists was highlighted, 
almost simultaneously with the appearance of Professor Chamberlin's 
and Mrs. Robinson's books, by Berle and Means' mammoth study of the 
modern corporation. 3 This study was also launched well prior to the 
Great Depression; its inspiration, Professor Berle stated in the preface, 
was the Wall Street boom and the attendant pyramiding of "industrial 
oligarchies." But it was not Professor Berle's interesting and erudite study 
of the changing property rights of the individual security holder but 
Gardiner C. Means' statistics on the industrial predominance of the na- 
tion's 200 largest non-financial corporations that captured popular atten- 
tion. These, he estimated, had combined assets at the beginning of 1930 
of $8 1 billion, or about half of all assets owned by corporations. Although 
open to challenge as to detail, his calculations buttressed his contention 
that "the principles of duopoly have become more important than those 
of free competition." 4 The book had a popular as well as academic audi- 
ence and the figure of "200" became a magic symbol in subsequent inves- 
tigations of economic power. 

The new market categories, plus the evidence of Berle and Means and 
of the scholar's own eyes as to the need for them, set the stage for the 

3 A. A. Berle and Gardiner C. Means, The Modem Corporation and Private Property 
(New York, 1934). 

4 Ibid., p. 45. In particular it was pointed out against Means that his list of corpora- 
tions included rail, power, and communications utilities where large scale was inevitable 
and also well recognized and where the area of private discretion had been circumscribed 
bv the state. 


revived interest in monopoly and its allied issues. One must also empha- 
size the esprit which Chamberlin's and Robinson's works gave to students 
of the field. Even though they substituted a new set of frustrations for 
the old ones, the new ones were welcome. It has been suggested that "the 
most revolutionary feature of the monopolistic competition theories [was] 
the unprecedented pace at which they conquered their audience." 5 
Neither Chamberlin nor Robinson was destroyed and redestroyed as 
was Keynes a few years later. Their most effective critic, Professor 
Schumpeter, centered his attacks not on the validity of their analysis per 
se but more generally on the notion that it much affected the assessment 
of capitalist reality. 6 Rarely in economics have ideas had such an enthusi- 
astic and uncritical welcome. 7 


The effect of the new market categories on empirical investigation and 
the search for policy was first evident in Arthur R. Burns' The Decline 
of Competition, 8 w 7 hich appeared in 1936. Professor Burns began with 
the hypothesis that "elements of monopoly . . . can no longer be re- 
garded as occasional and relatively unimportant aberrations from compe- 
tition. They are such an organic part of the industrial system that it is 
useless to hope that they can be removed by law . . ." 9 He thereupon set 
himself the task of determining how "the resulting imperfectly or monop- 
olistically competitive system" 10 works, the proper objectives of public 
policy concerning it and the means of achieving them. Professor Burns 
went some distance with the first part of his task, i.e., the description of 
markets and of the monopoly elements therein. His book is still an excel- 

5 Robert Triffin, Monopolistic Competition and General Equilibrium Theory (Cam- 
bridge, Mass., 1940), p. 17. 

6 Business Cycles (New York, 1939), p. 63 ff.; and in particular Capitalism, Socialism 
and Democracy, 2nd ed. (New York, 1946), p. 79 ff. 

7 Discussion of both the theory and theoretical developments following in its wake will 
be found elsewhere in this volume (pp. 6-9, 17, 20-22 above). It may be said, however, 
that both Chamberlin's and Robinson's books were, to a remarkable degree, less a begin- 
ning than a climax. Although the theory was the subject of exhaustive scrutiny and nu- 
merous refinements and corrections in detail, little was added in the way either of 
fundamental revision or noteworthy extension. This is not meant to detract in any way 
from the importance of such a contribution as Triffin's, just cited. This book provides an 
indispensable survey and critique of the whole development and has brought the work of 
Pareto and Stackelberg into focus with that of Chamberlin and Robinson. Triffin clarifies 
the notion of the industry by relating it to the system of external influences that bear on 
the equilibrium of the firm, but the basic theoretical system emerges largely intact. 

8 New York, 1936. 

9 Ibid., p. 3. 

10 hoc. cit. 


lent guide to techniques of trade-association control of prices, market 
sharing, price leadership, and the like. He was also among the first of 
many to discover that the assumptions of the new theory involved a series 
of traps for anyone who ventured recommendations on policy. In a tradi- 
tion common to both bourgeois and socialist theorists (e.g., Enrico Barone 
and more recently Oscar Lange 11 ) he accepted as ideal the level of em- 
ployment and use of resources under conditions of pure competition. 
Monopoly or oligopoly, and the associated techniques of market control 
obviously inhibited such ideal use. But having rejected pure competition 
as an impractical goal, Professor Burns was brought face to face with the 
only apparent alternative to accepting things as they are, namely, to 
make the State the agency for planning or at least improving upon the 
allocation of resources. He accepted the alternative; but having grasped 
this nettle, he found himself faced with the further task of detailing the 
criteria and techniques for state control of resource use. On this he made 
little progress. The frustration proved to be a recurring one, for the norm 
of pure competition, however valuable as an intellectual design or model, 
provides few practical clues to action vis-a-vis markets where the possi- 
bility of such competition is rejected. This is a technical matter. There is 
the further question whether the State, in a capitalist society and espe- 
cially in the United States, is a deus ex machina that can institute such 
comprehensive planning. Is the State, in a capitalist society, able, by a 
process of deliberate decision, to revise the basic constitution of the capi- 
talist economy itself? 

The new theory also had an influence on less cosmic lines of empirical 
study. Investigations of individual industries— the ubiquitous and useful 
"industry study'— were soon oriented toward the new market categories. 12 
The way was also opened for a much more realistic analysis of the rela- 
tion between competitive and monopolistic industries where the two are 
juxtaposed in the same market. This was particularly useful in dealing 
with the agricultural markets and deserves a special word. 

So long as the standard dichotomy in market analysis was between 
competition and monopoly, it was obviously difficult to differentiate the 
purely competitive markets, in which most unprocessed farm products 
are normally sold, from the oligopolistic markets in which they are nor- 
mally resold. So long as it was deemed inappropriate (as well as impo- 
lite) to characterize milk distributors or meat packers as monopolistic, 
they tended to be grouped with farmers as participants in a competitive 

J1 See Oscar Lange and Fred M. Taylor, On the Economic Theory of Socialis 
(Minneapolis, 1938). 

12 See pp. 142-149 below [Bain]. 



market. However, in 1934 Professor J. M. Cassels distinguished between 
the behavior or the monopolistic and competitive firm in analyzing the 
responses of farmers to the class prices established by a monopolistic milk 
marketing co-operative. 13 Much more extensive use of this analysis has 
been made by Professor W. H. Nicholls, first in research bulletins of the 
Iowa Agricultural Experiment Station and in periodical articles, and later 
and more elaborately in his book Imperfect Competition Within Agricul- 
tural Industries. 14 The latter, though formally presented as an analysis of 
agricultural markets, is also a theoretical totir de force of formidable pro- 
portions. After formulating the conditions of demand (specifically of de- 
rived demand) for farm products, the author elaborates the market solu- 
tions to be expected under various combinations of few and many sellers, 
few and many buyers, and of bilateral monopoly, and with varying as- 
sumptions as to the degrees of interdependence recognized by market 
participants. The study ranks with Trifrm's in the virtuosity with which 
theoretical tools are employed and it is a singularly useful guide to mod- 
ern market literature. It is also, as Professor Nicholls himself is at some 
pains to make clear, a good example of the frustration which accompa- 
nies efforts to develop a "theory" of oligopoly along lines of the conven- 
tional approach to the competitive market. Professor Nicholls makes a 
limited selection from an endless number of equally plausible assump- 
tions about individual behavior in the markets with which he is con- 
cerned. 15 He was eventually forced to conclude that the principal utility 
of such analysis in markets characterized by oligopoly or its counterparts 
is "to sharpen . . . thinking and tools of analysis in order to do a better 
job of empirical work." 16 It is not to be supposed, however, that this lim- 
ited achievement is without importance. Even in the absence of a 
''theory" of the oligopolistic market, attention was usefully focused on the 
diverse behavior of the competitive and monopoloid sectors of the econ- 
omy. This was of immediate assistance in explaining the different be- 
havior of agricultural and industrial prices during the 'thirties and 'forties 
and it at least established a framework for appraisal of relative levels of 
resource employment in agriculture and industry. The latter lines of in- 
vestigation rank with the most important of recent developments in agri- 
cultural economics. 17 

33 A Study of Fluid Milk Prices (Cambridge, Mass., 1934). 

11 Ames, 1 94 1. 

10 He quotes the apt observation of Professor Wassily Leontief that, when dealing with 
oligopoly, "the real issue is that of selecting an appropriate set of fundamental assump- 
tions." Journal of Political Economy, August 1936, XLIV, p. 544. 

16 W. H. Nicholls, op. cit., p. 165. 

17 They are most fully elaborated in T. W. Schultz. Agriculture in an Unstable Econ- 
omy (New York, 1945). 



The revival of interest in the theoretical aspects of monopoly and their 
application was paced, during the 'thirties, by a series of studies designed 
to measure the position of the large business unit in the economy. At least 
partly because of the extent and complexity of the statistical enterprise 
involved, much of this work was conducted under the auspices of the 
Federal Government. In scope and significance, it must be considered a 
landmark among government investigations. 

The inspiration to much of this work was Berle and Means' earlier 
calculations of the industrial predominance of the 200 largest corpora- 
tions. The most important of the further investigations was directed by 
Gardiner C. Means for the National Resources Planning Board. Making 
use of Bureau of Internal Revenue data, this study generally confirmed 
the pre-eminence of the large corporations. 18 Counting (subject to sortie 
possibilities of error) wholly controlled subsidiaries, the 200 largest non- 
financial corporations were credited for 1933 with control of between 19 
and 21 per cent of aggregate national wealth, between 46 and 50 per 
cent of the nation's "industrial" wealth, and approximately 60 per cent of 
the physical assets of all non-financial corporations. The study presented 
consolidated income and asset accounts for the 200 corporations and ex- 
amined their comparative predominance in different lines of economic 

The rationale of the study was an effort to discover the area of "ad- 
ministrative" as distinct from market co-ordination in the American 
economy. It was supplemented by an effort to establish the extent of the 
co-ordination between the 200 corporations (and an additional 50 finan- 
cial corporations) by means of interlocking directorates and "interest 
groupings." Interesting new ground was broken in the investigation of 
the interest groups associated with such nuclei as J. P. Morgan-First 
National, Rockefeller, Kuhn-Loeb, Mellon, Du Pont, and the Chicago, 
Cleveland, and Boston financial communities. 19 While the identifying 
bond between the interest groups ranged from such subjective factors as 
known working relationships or interlocking directorates to firm financial 
control, in the case of the Rockefellers and Du Ponts, the study cast use- 
ful light on the agglomerative tendencies at work within the handful of 
pre-eminent corporations. Of the 200 largest non-financial corporations 

18 The Structure of the American Economy, Part I, Basic Characteristics (Washing- 
ton, 1939), p. 99 ff. 

19 Ibid., p. 306 ff. This study was made by Paul M. Sweezy. 


and 50 largest financial corporations, 58 per cent of the industrial and 
utility assets, 82 per cent of the rails, and 51 per cent of the bank assets 
were aligned with one or another of the 8 interest groups mentioned 

Finally, in what was to become one of the most quoted researches of 
the decade, Dr. Means' group examined the extent of concentration by 
markets. Census industrial classifications, 276 in all, were ranked in ac- 
cordance with the proportion of the total output supplied by the largest 
four and largest eight firms. As a measure of market concentration, this 
exercise is crude. On occasion census classifications combine firms pro- 
ducing wholly unrelated products, while firms producing the same prod- 
uct may serve wholly distinct market areas. 20 Both defects, however, 
result in an understatement of market concentration, and the study was a 
convincing demonstration that oligopoly is the appropriate assumption in 
dealing with industrial markets in the United States. It went far toward 
establishing this assumption in American economic thought. 

The 200 corporations received yet another examination at the end of 
the decade in a study sponsored by the Securities and Exchange Com- 
mission for the Temporary National Economic Committee. As its title— 
The Distribution of Ownership in the 200 Largest N on- financial Corpo- 
rations—indicated, the focus was less on the economic power of the 200 
corporations, which was largely assumed, than on their ownership. 
Throughout the 'thirties, partly as the result of W. Z. Ripley's earlier 
polemics, 21 partly because of the work of Berle and Means, partly, no 
doubt, as a continuation of a much older concern over the ethics and con- 
sequences of absentee ownership, the locus of control over the modern 
corporation was actively discussed. Much of this discussion is beyond the 
scope of this essay, but it is noteworthy that by the end of the decade it 
was commonly assumed that in most large-scale corporate enterprise, the 
divorce of ownership from control, either in an immediate or an ultimate 
sense, was complete. Because of its susceptibility to shrinkage in reve- 
nues, pyramided stock ownerships through holding companies, especially 
of rail and electric power utilities, captured popular attention during the 
depression. The drive for corrective legislation and attendant investiga- 
tions made it clear to a large audience that the principal motive was ex- 
tension of control without ownership. At the same time there was an 
increasing tendency to view management, even where its position was 
unsupported by legal reinforcements, as a self-appointed trust (or, less 

20 Raymond W. Goldsmith, Rexford G. Parmelee, et al., T.N.E.C. Monograph 29 
(Washington, 1940). 

21 In particular his Main Street and Wall Street (Boston, 1939). 


politely, as a self-perpetuating bureaucracy) with a negligible ownership 
stake. Goldsmith and Parmelee, working principally with data obtained 
directly from the 200 corporations, showed that the stake of officers and 
directors in ownership was indeed comparatively thin and the holdings 
of officers especially so. Only about 6 per cent of the common stock of the 
200 corporations was owned by officers and directors; over half of all 
officers and directors had holdings that could be described as negligible. 
This could mean that corporate direction as exemplified by officers and 
boards had been largely professionalized but was still subject to control 
by principals. For at least a majority of the large corporations this might 
be the case, for the study showed the importance of family ownership 
cither through direct stock ownership or, more especially, through estates, 
trusts, and family holding companies. The authors observed that "a small 
group of dominant security holders is not in evidence in only 30 per cent 
of the 200 large corporations." 22 

Although Berle and Means had classified only about one-third of their 
200 corporations as "management controlled, " the emphasis they gave to 
this type of control, together with that exercised through legal devices or 
extreme minority holdings, left the impression that control of the modern 
large corporation by owners was exceptional. There is no assurance that 
this impression is unjustified. A sizable but quiescent ownership interest 
may allow an active and aggressive management to exercise complete and 
final control, and the possibility is not an unlikely one. Moreover, as a 
subsequent investigator has pointed out, 23 certain of the ownership inter- 
ests cited by Goldsmith and Parmelee are themselves owned by corpora- 
tions, a fact which in turn raises the question of the control of these 
corporations. In other cases the ownership was split between two or more 
dominant family groups with at least a presumption in some of these cases 
that management provided the decisive influence. 

Although the locus of ultimate power in the modern corporation re- 
mains, to a degree, conjectural, this element of conjecture might one day 
be removed for the two or three hundred larger corporations, by a de- 
tailed study of all of them. The issue is important and this would resolve 
it for business units responsible for the order of a third of all economic 
activity. One of the still unexploited opportunities offered by concentra- 
tion of control is that of abandoning generalization in favor of a com- 
plete apprehension of the universe. 

" 2 Goldsmith, Parmelee, et al., op. cit., p. xvi. 

23 R. A. Gordon, Business Leadership in the Large Corporation (Washington, 1945). 
Professor Gordon excluded twenty-four of the 200 corporations because a majority of 
the common stock was held by other corporations (in the case of three railroads) where 
there was control by lease by other corporations. 



It is now time to take the new ideas on monopoly into the world of 
affairs— to examine their nexus with depression and unemployment and 
the deeper questioning of capitalist institutions for which the depression 
provided a hospitable environment. At first glance, the new notions of 
generalized monopoly appeared to offer a ready explanation of contempo- 
rary distress. The classical solution of monopoly had always shown 
entrepreneurial returns maximized (or protected) at the expense of pro- 
duction. Under the label of imperfect competition, the monopoly solution 
could now be considered not the exception but the rule. It followed that 
restricted production and excess capacity were also the rule and their 
concomitant was unemployed resources. Not many accepted this vulgar 
formulation. Once outside the universities, however, the new theory was 
undoubtedly credited with diagnostic and even therapeutic values which 
it did not, in fact, possess. 

At a more sophisticated level, the new work did strike a blow at the 
concept of inherent order in capitalist behavior: the doctrines of imper- 
fect (and monopolistic) competition ran sharply counter to what Profes- 
sor Roll has called the "optimal distribution of resources prejudice." 24 

Under some circumstances this might have been revolutionary. No 
idea is more deeply rooted in non-socialist economics than that of a rule 
of competition where the controlling tendency is for resources to be em- 
ployed by firms and distributed between industries in such manner that 
they are combined with maximum efficiency into products that give max- 
imum satisfaction. To be sure, for half a century economics has been a 
kind of trial of wits between those who sought to perfect this doctrine of 
ultimate harmony and those who— citing inequality and its perpetration 
by inheritance, external economies, immobility of resources, and other 
inhibiting forces— sought to limit it. But the doctrine was only completely 
vulnerable at one point and that was where monopoly entered— the de- 
fenders and attackers entirely agreed that monopoly Qcum oligopoly) was 
deeply subversive of the competitive model. And, since oligopoly was 
stubbornly resistant to incorporation in a new system, at least by the old 
methods, it destroyed without leaving anything in its place. 

One or two scholars have seen the development in the foregoing light. 
The late Heinrich von Stackelberg, whose book Marktform und Gleich- 
gewichtr 5 was published almost simultaneously with those of Chamberlin 

24 Eric Roll, A History of Economic Thought (New York, 1942). 
~ 5 Vienna and Berlin, 1934. 


and Robinson, and who dealt extensively with the problem of oligopoly, 
seems as a result of his analysis to have abandoned all hope for an eco- 
nomic order except as provided by the State. Many have considered it 
relevant that he was the most prominent German economist to identify 
himself whole-heartedly with National Socialism. Professor Eduard 
Heiman has also expressed alarm at the implications of the theory. He 
suggests that "it is time to recognize that the concept of a system of mo- 
nopoly is self-contradictory and the very negation of everything econom- 
ics stands for." 20 Oddly enough, Professor Heiman makes no effort to 
deny the existence of a system of monopoly cum oligopoly and conse- 
quently comes close to enjoining economists to avoid thinking of the 
world as it is. 

Few American or British scholars drew any such nihilist conclusions. 
This was partly fortuitous. For many years prior to the revived interest in 
monopoly, work in the great tradition of Marshallian partial equilib- 
rium analysis had been carried on with little regard to larger issues. 
Although in most American universities it still occupied the area thai 
was honored with the label "economic theory/' many if not most of its 
practitioners had narrowed their interests to questions of product price 
and factor cost determination. Investigators of such alien subjects as 
business cycles, money and banking, and international trade had appro- 
priated the problems that are so painfully relevant to the real world. 

More important, during the years of depression economic theory 
tended increasingly to polarize on two distinct though not unrelated 
norms. The first was the goal of an appropriate employment of resources, 
the alternative being idleness; the second was the goal of an appropriate 
employment of resources, the alternative being (in some sense) a socially 
less efficient employment of resources. In the first instance, the problem 
of monopoly bore upon the question of resource use when the opportu- 
nity cost was less efficient employment. The depression, obviously, was 
focusing attention on the seemingly far more urgent question of any em- 
ployment vs. unemployment. This emphasis was enormously sharpened 
by the publication of Keynes' General Theory. 21 Although the assump- 
tion of imperfect competition is explicit in his analysis of the labor market 
and implicit in a good deal of his treatment of prices and of capital mar- 
kets, Keynes was largely oblivious to either the old or the new market 
categories. Moreover he treated the problem of relative efficiencies of 
employment with something between neglect and contempt: "There is 
no reason to suppose that the existing system seriously misemploys the 

-''History of Economic Doctrine (New York, 1945), p. 219. 

-''The General Theory of Employment, Interest and Money (New York, 1936). 


'actors of production which are in use." 28 There is no need to emphasize 
:he extent to which Keynes captured the attention of economists con- 
cerned with policy and herewith the interest that might otherwise have 
centered on the question of efficiency. 

In one sense the new theory made it even more difficult than hitherto 
to establish a relationship between monopoly and economic distress. 
Under conditions of pure competition, rigorously defined, cyclical insta- 
bility is conceptually possible. To monopoly and monopoloid forms, how- 
ever, one can trace disparity in income and both a lower and a less stable 
consumption function than would be expected under pure competition. 
Imperfect competition, also, can be conceived as breaking the connec- 
tion, through the interest rate, of the supply of savings and the demand 
for capital. It follows that imperfect competition or monopoloid forms are 
a necessary condition for consumption and investment fluctuation, and at 
least some types of inventory fluctuation. Thus imperfect competition or 
rather the absence of absolutely pure competition, including the labor 
market and the capital market, can be offered as a nearly comprehensive 
'cause" of cyclical fluctuations. But this is not to say very much. The 
appearance of oligopoly and its counterpart on the buyers' side of the 
market and monopolistic competition far antedated the theory that inter- 
preted them. And a most portentous concomitant of the new theory was 
that in making it easy to assume that monopoloid forms were general in 
the economy, it made it difficult to recommend their elimination as a 
reform measure. A diagnosis that had related the stagnation of the 
'thirties to monopoly in the old-fashioned sense would have made the life 
of any available monopolist miserable in the extreme. Once oligopoly and 
monopolistic competition entered the picture, to prescribe the elimination 
of monopoly became tantamount to demanding a wholesale revision of 
the economic order. Economists, some sections of the press oddly to the 
contrary, are not given to such violent prescriptions. 29 The highly re- 
stricted definition of pure competition, which the new theory brought into 
use, also helped make the competitive goal seem remote and impractical. 
In part, it should be added, this was the result of a too literal transference 

~ s Ibid., p. 379. 

20 It is intriguing that it is this revolutionary formula which Professor Hayek advances 
in its most uncompromising form. "The price system will fulfill [its] function only if 
competition prevails, that is, if the individual producer has to adapt himself to price 
changes and cannot control them." The Road to Serfdom (Chicago, 1944), p. 49. 
There is no nonsense here— the curve of the individual seller must be completely elastic 
at the ruling price. 

To effect such a reorganization of the economy would, I suspect, take some of the 
most formidable planning (and one of the largest bureaucracies) of all time. Before it 
was completed both Professor Hayek and his book would have alienated some of their 
most devoted admirers. 


of a scientific definition into the world of affairs. Without doubt there are 
many imperfect markets in which a "workable" competition yields the 
same effective solution, or has the same social effect, that the textbooks 
have associated with pure competition. 30 Apart from some not too conse- 
quential waste, I suggest that much of what is called monopolistic com- 
petition could be so classified. 

The most important effort to build a bridge between the new market 
categories and the theory that dealt with them, on the one side, and the 
depression, on the other, was by way of price behavior. Given widespread 
oligopoly and hence a large area of entrepreneurial discretion in the set- 
ting of prices, attention was naturally directed toward the way in which 
the discretion was exercised— and the criteria of private and social benefit 
by which it should be exercised. In the latter half of the 'thirties, the 
problem of price policy assumed the stature of a new field of economic 
investigation. In its social aspects it was taken up somewhat gingerly. The 
notion that a private firm could be guided by considerations other than 
its own short- or long-run interest was not one that all economists em- 
braced with appetite. To concede that a businessman should orient his 
price policy to social norms is to assert that the single-minded pursuit of 
profit is presumptively anti-social. It admits of a rule of private collec- 
tivism that accords important legislative functions to the private entre- 
preneur. Conservatives and liberals alike found the idea unappetizing. 
The implied alternative, namely that price behavior had become a fit area 
for state intervention, also had disagreeable overtones. 

However, the debate over price policy had less to do with such broad 
philosophical issues than with the much more concrete problem of the 
different patterns of price behavior during the depression and their effect 
in accentuating deflation. The differences in frequency and amplitude of 
price change between different price series were one of the most thor- 
oughly investigated phenomena of the 'thirties. (The reciprocal behavior 
has been equally apparent, though less thoroughly studied, during the 
postwar inflation.) Again the pioneer in these investigations was Gar- 
diner Means. 31 He measured the frequency and amplitude of movement 

30 A persuasive argument along these lines is contained in the manuscript drafts of 
Professor Corwin D. Edwards' forthcoming book on policy for maintaining competition. 

:il In The Structure of the American Economy, p. 122 ff. Means' efforts to measure 
concentration of economic power and associated phenomena were clearly among the 
important research achievements of the 'thirties. It is not clear that either the studies 
or author have won the recognition they deserve. This may possibly be explained by the 
rather novel framework into which Means fitted his work. Instead of bringing statistical 
measures to bear on the conceptual framework used by other workers, he had a tendency 
to create his own. And because his framework was unfamiliar and at times, perhaps, 
somewhat artificial, his work had less influence than might otherwise have been the case. 


of different series during the deflation phase and devised rough indexes 
of "depression sensitivity" for different commodities. He then related this 
sensitivity to concentration, type of product, and other characteristics of 
the industry. Both the conceptual and statistical aspects of the phenome- 
non of price flexibility and inflexibility were taken up by other students, 32 
culminating in a detailed survey of the whole issue for the Temporary 
National Economic Committee. 33 

Although unchanging or infrequent price changes of small magnitude 
are not inevitably a concomitant of imperfect competition under condi- 
tions of changing demand— a formal point that has been made with some 
vigor by Professor Scitovszky 34 — they are a possible concomitant, as they 
are not of pure competition. For a variety of reasons, infrequent price 
changes are a likely price solution under oligopoly. To avoid change is by 
all odds the simplest way of maintaining the oligopolistic entente to 
which the seller is a party. 35 

A number who observed the phenomenon of inflexible prices took an 
uncomplicated view of its cyclical effect. Assuming a rule of pure compe- 
tition, rigidly defined, the price dispersion associated with imperfect com- 
petition would not occur. Hence, whatever effect price dispersion might 
have in accentuating deflation was the result of imperfect competition. 
Hence the corollary: any steps that would diminish inflexibility would 
enhance cyclical stability. This primitive analysis of price inflexibility 
cum monopoly was extraordinarily influential in the making of actual 
policy during the 'thirties and it is not without influence today. 

To say that a flexible competitive economy has greater cyclical stability 
than an inflexible and monopolistic one is to say little that is useful. The 
real question is whether, given a rule of monopoly or monopoloid forms, 
stability is enhanced by increasing the area of competitive and flexible 
prices. Here the conclusion is a good deal less certain. During the 'thirties, 

32 See, in particular, Edward S. Mason, "Price Inflexibility," Review of Economic 
Statistics, May 1938, XX, pp. 53-64; Donald H. Humphrey, "The Nature and Meaning 
of Rigid Prices, 1 894-1 933," Journal of Political Economy, October 1937, XLV, pp. 
651-661; Donald H. Wallace, "Monopoly Prices and Depression," Explorations in 
Economics (Cambridge, Mass., 1936), p. 349 ff.; and my own paper "Monopoly Power 
and Price Rigidities," Quarterly Journal of Economics, May 1936, L, pp. 456-475. 

33 Saul Nelson and Walter G. Keim, T.N.E.C. Monograph 1 (Washington, 1941). 

34 Tibor de Scitovszky, "Prices Under Monopoly and Competition," Journal of Political 
Economy, October 1941, XLIX, pp. 663-686. 

35 Professor Oscar Lange observes in Price Flexibility and Employment (Bloomington, 
1944), pp. 86-87, that "the formation of monopolistic and monopsonistic group be- 
havior is not merely the result of greed for profit.' Rules of oligopolistic and oligopsonis- 
tic group behavior emerge because, without them, no firm would be able to predict the 
reaction of other firms to a change in its price." I would like to urge that the most 
elementary rule of behavior under oligopoly is to minimize the number of price changes 
and hence the number of times the understanding among oligopolists is put to a test. 


the dominant tendency, without doubt, was to look upon price dispersion 
and the accompanying alteration in the terms of exchange between 
groups as an accentuating force in the downswing of the cycle. And the 
stable rather than the cyclically flexible prices tended to be regarded as 
the active or disturbing factor. (It is interesting that in the reverse situa- 
tion of the 'forties, the flexible prices have been quite commonly viewed 
with alarm, the inflexible prices with some esteem.) The selection of the 
inflexible prices as the devil of the piece seems, however, to have been 
based more often on tradition than on analysis of demand and income 
effects. On the other hand there has been a strong post-Keynesian tend- 
ency to regard the cyclically inflexible prices as a stabilizing influence in 
the cycle. The most detailed argument has been advanced by Professor 
Hansen. 36 Associating cyclical fluctuations with changes in income re- 
sulting primarily from changes in investment activity, he argues that 
price dispersion is a symptom but not a cause of deflation and that "cycli- 
cal price flexibility all around ... at the end of a boom, might well 
accelerate the downswing." 37 He does place emphasis on the importance 
of structural price flexibility— the adjustment of prices to changes in unit 
costs— a distinction which has been taken up by others. (It may be noted 
that to the extent that both cyclical and structural inflexibility are a con- 
comitant of monopoly power, as they undoubtedly are, it may be difficult 
in practice to have one without getting the other.) Professor Hicks has 
also argued that stable prices, and by implication the imperfect competi- 
tion with which they are associated, may act as a stabilizing influence in 
the economy. 38 A not dissimilar conclusion is reached by Professor Lange 
in his elaborate theoretical treatment of the relation of price behavior to 
economic activity. 39 He discards price flexibility as a norm for the modern 
oligopolistic economy— he does not deny that it may have served as a 
stabilizing influence in past periods— and argues that areas of rigid prices 
will minimize the amount of monetary management and public expendi- 
ture necessary to check the downswing of the cycle. 

The relation of differential price behavior to the cycle or, more broadly, 
to the level and stability of resource employment is clearly unfinished 

30 Cf. Fiscal Policy and Business Cycles (New York, 1941), p. 313 ff., which closely 
follows his essay in The Structure of the American Economy, Part II, National Re- 
sources Planning Board (Washington, 1940). 

37 Fiscal Policy and Business Cycles, p. 322. Professor Hansen does not deny the use- 
fulness of cost-price adjustments in other phases of the cycle, although he insists that 
they are of subsidiary importance. 

88 Value and Capital (Oxford, 1939), pp. 265-271. Hicks argues that rigid prices of 
factors are certain to be stabilizing where those not offered at the given price remain 

3U Op. cit., p. 87 ff. 


business. Discussion remained active until war brought the overriding 
agreement that a high and highly organized output required stable prices 
and made the issue temporarily irrelevant. 

In academic circles, price behavior and policy proved to be the most 
durable facet of the revived interest in monopoly. Well before the out- 
break of the war, interest in academic circles in the theoretical issues 
associated with monopoly and competition was on the wane. At that time 
Professor Roll suggested, rather tentatively, that this lull occurred be- 
cause the theoretical possibilities of the subject "are now exhausted" and 
the field was being abandoned to "descriptive" economists and those con- 
cerned with policy. 40 Be that as it may, the decline in interest in the uni- 
versities was followed by a great burgeoning of interest in Washington. 
To that I now turn. 


It has often been suggested that the "New Deal" lacked any defined 
economic "philosophy." Nearly the reverse is true— it had several of them. 
President Roosevelt, it now seems clear, was singularly uncommitted to 
any particular economic dogma within the broad framework of a liberal 
capitalist faith. This, however, made it possible to win his support for 
any persuasively argued idea. That opportunity was considerably ex- 

The first few years of the new New Deal were a Dutch pie that con- 
tained a small present for everyone— budget orthodoxy for the proponents 
of sound finance, currency and exchange manipulation for the monetary 
enthusiasts, price-fixing for those who were being punished too severely 
by deflation, a good deal of vague "planning" for those who had attended 
too many seminars in political science. Gradually two positions on eco- 
nomic policy began to dominate the others. One can sufficiently and not 
inaccurately be called the Keynesian view. The other had its citadel in 
the Department of Justice and its outposts in the Securities and Ex- 
change Commission, the Federal Trade Commission, and, very impor- 
tantly, in the Congress. It saw monopoly and the concentration and abuse 
of economic power as the principal problem. In general those who em- 
phasized fiscal and monetary solutions were principally concerned with 
the immediate issues of income and employment; the anti-monopolists, 
on the other hand, regarded themselves as the architects of a permanent 
reform of capitalist institutions. Although this difference helped sustain 
a reasonably amiable coalition between the two groups, its importance 

40 Eric Roll, of. cit., p. 523. 


should not be exaggerated. For the enthusiasts, both were complete the- 
ologies wholly capable of answering the most urgent economic questions 
of the day. 

Not since the time of Theodore Roosevelt— perhaps not even then- 
had there been so much interest in the anti-trust laws as in the late 
'thirties, or such energetic efforts to enforce them. Although this activity 
was nourished by the contemporary academic interest in monopoly and 
imperfect competition, its most important roots were elsewhere. The 
principal entrepreneurs were lawyers. In their system of economic theory 
competition was good and performed indispensable regulatory functions. 
Monopoly was evil. So for many, who were in the Brandeis tradition, 
was size. The sophisticated problems which modern theory had associ- 
ated with these concepts were not troublesome. For men who, in an hon- 
orable tradition of the American bar, were expiating a lifetime of service 
to American corporations by spending a few years harassing them, the 
suppression of formal collusion or deliberate conspiracy was, ordinarily, a 
sufficient goal in itself. 

The anti-monopoly drive of the 'thirties drew strength from other 
sources. It was highly agreeable to progressives— including liberal con- 
gressmen—who, in a world of new and dubious formulas and amid 
charges and countercharges of sinister ideology, found comfort in the 
tried and tested radicalism of the Grangers and Populists. It even at- 
tracted a measure of support from businessmen and the conservative 
press. In a world where it was hard to defend the status quo, both were 
inclined to pay grudging lip service to what at least seemed like an effort 
to restore the status quo ante. 

Most important of all, the anti-monopoly drive of the late 'thirties drew 
strength from the remarkable energy and personality of its leader, Thur- 
man Arnold. Arnold was something of a convert. His The Folklore of 
Capitalism, 41 published in 1937, deprecated the anti-trust laws— he went 
so far as to assert that they were a fagade which protected large corpora- 
tions from more effective regulation. He suggested that "Theodore 
Roosevelt never accomplished anything with his trust busting. Of course 
he didn't. The crusade was not a practical one," and observed of past 
years that "whenever anyone demanded practical regulation, they [the 
anti-trust laws] formed an effective moral obstacle since all liberals 
would answer with a demand that the anti-trust laws be enforced." 42 On 
coming to Washington as head of the Anti-Trust Division of the Depart- 
ment of Justice, Arnold resolved his doubts. He launched a program not 

41 New Haven, 1937. 

42 Ibid., pp. 211, 217. 


only of vigorous enforcement of the laws but of equally vigorous claims 
on their behalf. 

The "Arnold era" lasted, roughly, from 1938 to 1941. Arnold obtained 
substantial increases in the appropriations for anti-trust enforcement— 
from $435,000 in 1936 to $1,325,000 in 1941— and recruited an excep- 
tionally able and spirited force of subordinates. His activities were char- 
acterized by both imagination and a well developed sense of drama. In 
addition to such conventional targets as oil companies, ALCOA, and the 
glass container industry, he reached out to such unsuspected (and un- 
suspecting) offenders as the American Medical Association and the As- 
sociated Press. A feature of his enforcement were the drives on entire 
industries, of which the most notable was that in 1940 on the building 
trades. For the first time anti-trust action was brought comprehensively 
to bear on a great number of local guild-monopolies which, previously, 
had enjoyed substantial immunity as the result of their small scale. This 
drive was also distinguished by the inclusion of the unions— once a fa- 
vored target of the Sherman Law enforcement— although eventually (in 
the Hutchison Case) they were held to be substantially immune to such 

During these years the anti-trust laws could perhaps be said to have 
had a fair trial. It was not an extended trial; it was subject to most of the 
numerous administrative and procedural handicaps that have always 
plagued anti-trust enforcement. 43 Yet, while one could reasonably ask for 
enforcement that was always as vigorous as during these years, it is 
doubtful, as laws are enforced in the United States, if one could ask for 
much more. What is to be concluded from this experiment? 

It is easy— perhaps too easy— to say what the anti-trust laws cannot do. 
Certainly they cannot positively alter the basic structure of a capitalist 
economy. There is no evidence that the ownership and control of Ameri- 
can industries was any less concentrated in 1940 than in 1935 or tnat ^ 
would have been more than marginally different by 1950 had the drive 
continued in the tempo of the late 'thirties. Nor is there any evidence 
that the drive contributed to recovery or any reason for supposing that it 
was capable of making the economy less subject to cyclical instability. In 
his requiem on the Arnold era 44 Thurman Arnold does make this case, 
but only by assertion. Further, it is now clear that anti-trust enforcement 
has, at best, only a tenuous connection with the factors which are signifi- 
cant in the monopoly or oligopoly equilibrium. As Professor Edward S. 

43 See Professor Walton Hamilton's colorful monograph for the T.N.E.C., Anti-trust 
Action, T.N.E.C. Monograph 16 (Washington, 1941). 
^Bottlenecks of Business (New York, 1940), p. 12 ff. 


Mason 45 and others have observed, there is a notable gap between the 
legal and economic concepts of monopoly. The first tends to emphasize 
behavior in the market; the second is naturally concerned with the result 
or solution that is achieved. Anti-trust enforcement can, at best, make 
contact with only a few of the types of behavior— and not necessarily the 
most important ones— that are capable of yielding a monopolistic solu- 
tion. 46 

The final problem of anti-trust policy, and the one which the theoreti- 
cal work of the 'thirties made peculiarly evident, is its inability to make 
satisfactory contact with oligopoly. It is quite possible that imperfect 
knowledge, inertia, and the allowance that is made for public opinion all 
help make the market solution under oligopoly indistinguishable from 
that under monopoly. In any case there is, a friori, no reason for suppos- 
ing that oligopoly has social consequence that is inherently more benefi- 
cent than single-firm monopoly. But oligopoly cannot be like competition. 
The oligopolist cannot escape from the circumstances that vest him with 
the power to influence the common market, and the formal solutions that 
represent him as ignoring the market effects of his actions are barren 
novelties. If this is granted, then it is just as important that the anti-trust 
laws come to grips with oligopoly as with monopoly. Or more so, since, 
from the statistics of concentration noted above, it is clear that oligopoly 
must be counted the ruling form in industrial markets in the United 
States. Oligopolistic price policies, even though they give results similar 
to those of pure monopoly, are presumably immune from anti-trust inter- 
ference so long as no express or tacit collusion among sellers can be 

The problem would hardly be solved were the institution of oligopoly 
brought within the scope of anti-trust action either by legislation or judi- 
cial interpretation. In the past, the courts have shown themselves willing 
to punish those whose guilt has been established under the Sherman Act 
although the penalties have frequently been less than drastic. Especially 

45 "Monopoly in Law and Economics," Yale haw journal, 193,7, XL VII, p. 37. 

40 A recent writer, Professor Rostow, has taken considerable comfort from two recent 
decisions which, he holds, markedly narrow this gap between monopoly in law and in 
economics. In the case of the Aluminum Company, Judge Learned Hand held that 
market dominance, regardless of intent, forced ALCOA to behave as a monopolist; and 
the Supreme Court recently took cognizance of non-collusive oligopolistic behavior by 
the big tobacco companies. While these cases are not without importance, it can hardly 
be argued that they are more than a minor break with a tradition that is strongly 
behavioristic and which, in view of the difficulty of distinguishing ideal from unsatisfac- 
tory market solutions, is perhaps necessarily so. Cf. Eugene V. Rostow, A National Policy 
for the Oil Industry (New Haven, 1948), p. 123 ff. The decisions are those of Judge 
Learned Hand in the Aluminum Case (U.S. vs. Aluminum Company, Circuit Court 
of Appeals, 2d, 1945) and of the Supreme Court in American Tobacco Company vs. U.S. 
(328, U.S. 781, 1946). 


under Arnold's leadership, the Anti-Trust Division had considerable suc- 
cess through consent procedure, in banning overt collusion in restraint of 
trade. These, and other remedies, all leave the basic market structure 
unchanged and the Courts have shown the utmost unwillingness to in- 
voke remedies that do involve extensive structural changes. 47 Yet for 
oligopoly as for monopoly there is no other remedy that strikes at funda- 
mentals. For monopoly, dissolution can indeed be regarded as a remedy, 
inasmuch as all customary definitions or concepts w 7 ould confine it to a 
relatively limited number of firms. Such a remedy for oligopoly implies 
application to a large part of the economy. No one with a sense of history 
could suppose that the Courts would or could contemplate such a whole- 
sale reorganization of the economy. 

Nevertheless the anti-trust laws remain a useful instrument of social 
control and, quite possibly, the "Arnold era" showed how useful they can 
be. Although anti-trust action cannot produce important structural changes 
in the economy or even, in a negative sense, much retard basic trends 
toward corporate growth and concentration, it still may serve to "im- 
prove" the ruling equilibrium. The danger of anti-trust prosecution, or of 
public ill will leading to intervention by the Department of Justice is, 
without doubt, a fairly important consideration in corporate price policy. 
It is fair to assume, as a result, that prices are set closer to marginal costs 
—i.e., there is a lesser degree of monopoly in the economy— than would 
otherwise be the case. This, presumptively, means higher income and 
output and greater satisfactions than otherwise. 48 Without doubt Thur- 
man Arnold was the ghost at more conferences on corporate price policy 
than either his predecessors or successors in office and, accordingly, the 
effect of the anti-trust laws on the oligopoly equilibrium was correspond- 
ingly more beneficial during his tenure. 

Finally no one should underestimate the importance of the anti-trust 
laws in bringing business practice into accord with basic concepts of de- 
cency and equity or in preventing those with economic power from using 
it to combat innovation. It is not true that monopoly or oligopoly always 
breeds senescence and protection of the status quo, and the reverse may 
often be the case. The monopolistic or oligopolistic firm is likely to 
spend more money for research, if for no other reason than because it is 

47 The recent divorce of Pullman car operation from car manufacture is principally 
noteworthy as an exception to the rule. In the recent decision in the Aluminum Case, 
although the Court found that ALCOA had monopolized the ingot market and that, 
in writing the Sherman Act, the Congress "did not condone 'good trusts' and condemn 
'bad' ones, but forbade all," it contented itself with postponing action until the effect 
of sale of government-owned war plants could be foreseen. 

48 Cf. Professor Hansen's plea (op. cit.*) for "structural price flexibility," which comes 
to the same thjnsx 


likely to have more to spend. In addition, because the firm is large in rela- 
tion to the total industry, its share in the market for (say) a new product 
will remain considerable even though the innovation is appropriated by 
the industry at large and there is more than a chance that the benefits 
from cost reduction, though similarly generalized, will be perpetuated 
in the new oligopolistic equilibrium. None of these conditions hold under 
conditions approaching pure competition. It is interesting, not as proof 
but as illustration, to compare the rate of innovation in the oil industry 
with that in the bituminous coal industry. In the former, innovation has 
been richly financed and rapid. In the bituminous coal industry, which 
approximates conditions of pure competition, there is little research and 
little progress. It is also hard to see how any considerable research ex- 
penditure could be advantageous for most individual mine operators- 
even assuming that year in and year out, they could afford it. One should 
notice, also, the case of agriculture, where, as an aspect of pure competi- 
tion, virtually all research is necessarily conducted by the government. 
But this is not a field for easy generalization. In supporting Group Health 
against the American Medical Association, Arnold showed that the anti- 
trust laws were both an important and effective weapon on the side of 
innovation. This was also shown in his attack on the restrictive covenants 
and conventions of local builders and building trades. 

In placing the anti-trust laws in perspective, one has to conclude that 
they are not serviceable for many of the cosmic purposes that their ardent 
proponents hold sacred. But it would also be unfortunate if one were to 
seem to prove too much. 


The anti-monopoly crusade came to an end with the war. A number 
of business executives who came to Washington to help arm the republic 
felt that business would not be able to give war production its undivided 
attention until Thurman Arnold was safely leashed. With the help of the 
Services they devoted themselves unselfishly to that task. At a time when 
competition was being set aside and capitalism substantially adapted to 
planned production of war goods it was inevitable that they should suc- 
ceed. Meanwhile, interest in Washington had partly shifted from Thur- 
man Arnold's indictments to yet another manifestation of the revived 
interest in monopoly. This was the Temporary National Economic Com- 

The T.N.E.C. was billed as a catholic examination of modern capital- 
ism; and its genesis, at least in part, was in the recession of 1937. 
However, in the main it was, as the public called it, a "monopoly" inves- 


tigation; the stage for it was set by Arnold's spectacular drives, by the 
preoccupation of a keen group of legal scholars in the Securities and 
Exchange Commission and elsewhere with corporate size and the philos- 
ophy of regulation, and, more remotely, by the academic discussion of 
imperfect competition. More important, perhaps, as immediate back- 
ground, were Gardiner Means' investigations of economic concentra- 
tion and concomitant price behavior. These had been extensively dis- 
cussed in Washington and were cited by the President in his message 
(April 29, 1938) requesting the investigation. The message also sug- 
gested that unemployment was the product of an inflexible price struc- 

The T.N.E.C. was in a great Anglo-American tradition of conjoined 
lay and expert inquiry into economic questions. It was carefully planned. 
Funds were reasonably adequate. It had wide access to economic data at 
a time when these had reached a high order of excellence. It reached out 
to command a large amount of specialized talent. Yet by almost any 
standards the T.N.E.C. was an undistinguished and disappointing en- 

The Committee amassed a great deal of information about the Ameri- 
can economy. Unfortunately, information that is not purposefully organ- 
ized is a depreciated currency, and much of the T.N.E.C.'s contribution 
could be so described. Much of it lacked even novelty. The investigation 
of the steel industry went laboriously into the basing point system. It 
would seem incredible, at this stage in American history, were anything 
added to knowledge of this venerable institution, and nothing was. The 
investigation of the inflexibility of steel prices was more interesting, 
although it served to show that organized adversary debate between 
economists is poor scientific method. Elsewhere— on industrial insurance, 
the role of insurance companies in the concentration of economic power, 
the disenfranchisement of policy holders in mutual insurance companies, 
patent monopoly, and other subjects— the Committee broke into some 
new territory. But even on such subjects as insurance, where the find- 
ings were not without importance, they were rather by the way of con- 
firming long-held opinions not only of economists but of the public at 


49 As frequently with such enterprises, the usefulness of the T.N.E.C. findings is 
partly defeated by their sheer bulk— the printed record itself runs to some 17,000 pages. 
Mr. David Lynch, in The Concentration of Economic Power (New York, 1946), has 
performed the useful task of summarizing the principal findings. The book is marred, 
however, by the author's rather mechanical approach to economic judgments and by his 
tendency to interpose in his assessment of T.N.E.C. findings his own overdeveloped 
sense of business evil. 


The work of the T.N.E.C. was somewhat retrieved by a series of use- 
ful monographs prepared under its sponsorship some of which have 
already been cited. 50 Mention should also be made of Professor Clair 
Wilcox's monograph, Competition and Monopoly in American Indus- 
try, 51 an ambitious and, on the whole, successful effort to classify a large 
sector of American industry in accordance with its market behavior. To 
do this Professor Wilcox established a certain number of definite 
categories of market control— simple monopoly, duopoly, monopoly 
through price leadership, through patent control, through market shar- 
ing, and so forth— and fitted (or on occasion crammed) industries into 
their appropriate categories. Under the direction of Willard L. Thorp 
and Walter F. Crowder, a further excursion was made into the statistics 
of concentration. In an attempt to measure trends in concentration, two 
indexes were constructed showing, on a 1 9 1 4 base, the number of estab- 
lishments accounting for half the wage earners in each Census industry 
and the proportion of all establishments needed to account for half of 
the workers. Although the effort was partly defeated by the relatively 
small number of years (eight in all) for which Census data were avail- 
able, and by adventitious factors which warped the data in certain of 
these years, the study did show a perceptible though by no means power- 
ful trend toward increased concentration— the present pattern of con- 
centration was approximately achieved prior to World War I. The 
monograph also examined concentration in individual industries and 
products as well as certain characteristics of administration of large-scale 
enterprises. 52 Among other monographs deserving of at least passing 
mention were A. C. Hoffman's on large-scale organization in the food 
industries, 53 a study somewhat paralleling the work of Nicholls already 
mentioned 54 ; the study by Helene Granby, Raymond Goldsmith, and 
Rexford Parmelee of security ownership in listed corporations 55 ; and the 
work of Marshall E. Dimock and Howard K. Hyde on Bureaucracy and 
Trusteeship in Large Corporations.™ 

If, the monographs aside, the investigatory part of the Committee's 
work was disappointing, it stands as a superb achievement compared 

50 Cf. pp. 107, 113, 117 above. 

51 T.N.E.C. Monograph 21 (Washington, 1941). 

52 Tfoe Structure of Industry, T.N.E.C. Monograph 27 (Washington, 1941). 

53 Large Scale Organization in the Food Industries, T.N.E.C. Monograph 35 (Wash- 
ington, 1 941). 

M Cf. p. 105 above. 

55 Survey of Shareholdings in lyio Corporations with Securities Listed on a National 
Exchange, T.N.E.C. Monograph 30 (Washington, 1941). 
50 T.N.E.C. Monograph 11 (Washington, 1941). 


with the Committee's interpretation of its findings and its recommenda- 
tions. The latter will be read avidly but only by the connoisseur or 
bromides. No serious effort was made to provide an appreciation or 
rationale of large-scale enterprise and concentrated economic power as 
facts of contemporary economic life. There was no effort to explain the 
malbehavior of the economy during the whole of the preceding decade. 
Indeed, there was no diagnosis of any kind. Rather, with a droll faith 
in some occult process of democracy, it turned the task over to the Ameri- 
can people. "The members of the Committee are not rash enough to 
believe that they can lay down a program which will solve the great 
problems that beset the world, but they are convinced that the infor- 
mation which this Committee has assembled . . . will enable the people 
of America to know what must be done if human freedom is to be 
preserved." 57 

The recommendations were as futile as the diagnosis. They called for 
faith in free enterprise, vigorous enforcement of the anti-trust laws, com- 
pulsory licensing and some miscellaneous patent reforms, registration of 
trade associations, an approach to federal charters for national corpo- 
rations, more business research, and better food, housing, and health for 
the underprivileged. Although no mention was made of the importance 
of regular prayer, Representative Summers did remind the Committee 
"that there is a living God whose laws control everywhere ... as 
distinguished from being governed by the theories of men." 58 In line, 
presumably, with his faith in divine ordinance, he dissented from a 
recommendation for the repeal of the Miller-Tydings Act. 

The reasons for the failure of the T.N.E.C. are not simple. Perhaps 
it would have done better had it had a more precise focus. Had it been 
avowedly a committee on monopoly and the concentration of economic 
power, it could hardly have avoided facing up to their implications for 
modern capitalism. The broader charter, though in principle desirable, 
allowed the Committee to sample too widely and too diffusely among 
issues relevant to the overriding issue of underproduction and unemploy- 

Also the T.N.E.C. came late in a liberal administration, at a time 
when it was less important to display crusading fervor than to demon- 
strate comparative respectability. Hearings of the T.N.E.C. were marked 

57 Final Report and Recommendations, T.N.E.C. Document 35 (Washington, 1941). 

58 Ibid., p. 50. Leon Henderson and Isador Lubin, the two economists most promi- 
nently associated with the work of the Committee in its final stages, declared the recom- 
mendations wholly inadequate. They cannot be criticized for failing to do more, for both 
at the time were preoccupied with defense activities. 


by none of the elan or sense of adventure of the Pecora investigations, 
for example. Witnesses and the public were assured, repeatedly, that 
it wasn't that kind of investigation. Unfortunately, there may be no 
other kind. Information that isn't in the public domain is usually being 
held out for a reason. 

At times the Committee, in its desire to display its respectability, 
verged on Philistinism. In its final report it characterized the case for a 
secular decline in investment as "un-American." The merits of that case 
are not at issue here; the Committee could quite properly dissent from 
an argument that it did not find convincing. It must be condemned, 
however, for what was clearly an attempt to stigmatize an important and 
well-reasoned point of view. 

There was a deeper reason for the failure of the T.N.E.C. The men 
who principally supplied its intellectual guidance were deeply committed 
to an ideal. That ideal was an economy in which the dynamic as well 
as the regulatory power was supplied by the competition of independent 
and comparatively small business units. But the Committee found itself 
exploring a world in which the typical industrial market is pre-empted 
by three, four, or half a dozen giant firms with, usually, a fringe of small 
hangers-on. There was no possibility of reconciling the ideal world with 
the real world. The Committee was too conscious of political reality to 
recommend (or even explore as a possibility) what would amount to a 
planned assault on the whole structure of modern corporate enterprise. 
And to replace giantism and oligopoly with modest-scale competitive 
enterprise would take nothing less. On the other hand the Committee 
was far too deeply committed to the small-scale, competitive ideal to 
admit of and prescribe for a monopolistic economy. Unable either to 
recommend what it wanted or to accept what it had, the Committee 
took the only available course. After declaring its faith in the anti-trust 
laws, an act of piety that never fails to sanctify the failure of a liberal's 
imagination, it submitted not a finding but an apologia, and quit. 


The war years were by no means barren of achievement, though little 
of it is of a sort that can be recorded in an essay of this kind. Economists 
assumed a commanding role in the design and administration of pro- 
duction, price, and distribution controls and in such diverse enterprises as 
wage stabilization, procurement, military intelligence, and international 
relations. Although there was a corresponding hiatus in scholarly pro- 
duction, it seems certain that the war experience will add realism and 


catholicity to economic research for many years to come. As a by-product 
of wartime economic planning there was also an important accretion 
of information on the American economy. Though, as noted, the problems 
of capitalism are not traceable to shortages of crude fact, the work of 
scholars will be both improved and made more difficult by the countless 
industry studies and market and wage analyses that were made by or on 
behalf of the war agencies. Even more important, the adaptation of the 
economy to war production threw unparalleled light on the mechanism 
that was being adapted. It will be unfortunate if the war experience is 
not carefully reviewed with this latter opportunity in mind. 59 

Two lines of inquiry relevant to this essay were stimulated by the war. 
The first concerned small business. For mobilizing economic resources 
a few large units are, without doubt, both more convenient and more 
effective than numerous smaller ones. During the war, capitalism was 
temporarily collectivized by the government. The great areas of private 
collectivism were brought within this system with comparative ease; 
numerous small units presented a far more serious problem. One result 
was that large concerns participated promptly and profitably in war 
orders; small concerns, on the other hand, were threatened with the loss 
of their labor or raw materials or of their markets. Because of incomplete 
mobilization and the compensating effects of increased income, the actual 
dangers to small business during the war were always more potential 
than real— the mortality rates for small enterprises were extraordinarily 
low. Nevertheless there were a succession of small business "crises" in 
Washington, and two congressional committees conducted semi-con- 
tinuous investigations of the problems of small businessmen and of 
sundry wolves in small businessmen's clothing. As the war progressed to 
be vocally sympathetic with small business became one of the most popu- 
lar manifestations of a social conscience. A probable result was that small 
plants received prime or subcontracts they otherwise would not have ob- 
tained and more attention than otherwise was given the small trader in 
the framing of price and rationing regulations, The chief harvest, how- 
ever, was in oratory. 

The war also focused attention on the large industrial combines of 
Germany and Japan and on the somewhat related question of inter- 
national cartels. At the end of the war the Japanese combines were the 
subject of a special study by a staff headed by Professor Corwin Edwards 

59 1 venture to refer to two papers of my own that were written with this end in view : 
"Reflections on Price Control," Quarterly Journal of Economics, August 1946, LX, pp. 
475-489; and "The Disequilibrium System," American Economic Review, June 1947, 
XXXVII, pp. 287-302. 


which, broadly speaking, recommended the retirement of the old families 
from both ownership and control, dissolution of the holding company 
structure, the elimination of banking control over industrial corporations, 
prohibition of control by single interest groups over unrelated activities, 
and protection of these reforms through enactment of an anti-trust 
statute. Special government machinery was outlined for the acquisition 
of securities from the family holding companies of the Zaibatsu with a 
view to resale to a more widely dispersed ownership. Subject to some 
modifications by the occupying authorities and a considerable lack of 
enthusiasm by the Japanese and some branches of the United States Gov- 
ernment, the foregoing has become stated policy in Japan. 

Policy in Germany has been confused by the fact of four-power occu- 
pation and by some marked extremes of American economic policy. 
The latter, at times, has amounted to arbitrary hostility to size qua size. 
The number of employees of a firm has been regularly advanced as a 
prime criterion of whether or not it should be split up; to divide any 
given firm into two or more parts has on occasion seemed a sufficient 
aim of policy. At times public ownership, in many cases a fairly obvious 
solution, has been resisted, not alone on usual conservative grounds, but 
because of a fear of "publicly owned cartels. " Although the American 
"decartelization" program has been urged with crusading zeal, little has 
been accomplished. There was considerable passive resistance within 
Military Government. The British also argued vigorously against the 
oversimplification of the norm of free enterprise and competition im- 
plicit in the American proposals. The French have supported the Ameri- 
can proposals, though apparently less on grounds of principle than from 
the conviction that they would make life unpleasant for the Germans. 
Initial support also came from the Russians, though the reasons would 
not be readily apparent from Marxian theory. 

Much, though not all, of the war and postwar interest in cartels grew 
out of the anchor position of the German combines in numerous prewar 
cartel arrangements. Part of this discussion has been rather romantic. 
German cartel participants have variously been credited with maintain- 
ing expert espionage networks, arming Germany and disarming her 
potential enemies, and, at the appropriate time, calling the signals for 
war. Peace and prosperity have been seen as largely contingent on the 
destruction of cartels. 60 At an adequately restrained level, the discussion 
has resulted in a useful history of cartel arrangements in sugar, rubber, 

00 The Department of Justice has been the source of much of this colorful analysis. 
Cf. Joseph Borkin and Charles A. Welsh, Germany's Master Plan (New York, 1943); 
and Wendell Berge, Cartels, Challenge to a Free World (Washington, 1944). 


nitrogen, steel, aluminum, and other products by Professor George W. 
Stocking and Dr. Myron W. Watkins 61 and a dispassionate survey of 
cartel and commodity agreement policy by Professor Edward S. Mason. 62 


The time has come for a brief word of summary. Quite clearly the 
last fifteen years have been marked by an active effort to resolve the 
problems presented by large-scale or monopolistic enterprise and to devise 
a public policy appropriate to their existence. It is apparent that, although 
the increment of knowledge has been considerable, both tasks have been 
attended by considerable frustration. The analytical task would appear 
to have failed because oligopoly, by all evidence the ruling market form 
in the modern economy, has not yielded to the kit of tools long em- 
ployed for analysis of the competitive market. In the competitive market, 
the inability of the individual to affect the solution made it possible to 
eliminate the vagaries of individual behavior from among the market 
data. It was possible to proceed, therefore, with a relatively simple set 
of assumptions. It is of the essence of the oligopoly solution that any 
individual can affect the solution. The analysis, therefore, had to take on 
a wholly unmanageable burden of assumptions as to how each participant 
in the market would behave. The whole exercise, as a result, bogged 

The dilemma in the field of policy is not unrelated. The problem 
of monopoly policy has long been intellectual property of men whose 
faith is in competition. A rule of oligopoly poses, for them, the unat- 
tractive alternatives either of recommending a wholesale dissolution 
of existing business units or of devising rules of behavior for a kind of 
society which none likes, which for some is a positive anathema, and 
to which conventional modes of analysis and thought are inapplicable. 

Happily this is not the place where such riddles have to be solved. 
But a suggestion is in order. The dilemma may be more intellectual than 
real. We do live in an industrial community where oligopoly— or, more 
horrid word, private collectivism— is the rule. But, strangely, we do live. 
Our dissatisfaction with our world is less the result of having known 

91 Cartels in Action (New York, 1946). 

62 Controlling World Trade (New York, 1946). The study was sponsored by the 
Committee on Economic Development. A special virtue of Professor Mason's study is 
that he approaches his task with an excellent sense of the relation of cartels to the prob- 
lem of equilibrium that arises when a small number of large sellers are juxtaposed in 
limited markets. It is this, rather than man's propensity for evil, that brings cartels into 


any other than of having constructed a model of another economic 
society, the rationale of which we know and which is more companion- 
able to our sense of elegance and order. We shall never find anything so 
agreeable in the world we have. But perhaps there will be compensation, 
once we have exchanged elegance for actuality, in a greater rate of 
progress in understanding what we have. 


Joe S. Bain 

The character and consequences of the price and output decisions of 
business firms in various industries have been studied for many years 
by American economists. Since the early 1930's, however, these matters 
have been the subject of a somewhat more intensive and systematic 
study, and a fairly well-defined field, bearing the label of "Price and 
Production Policies" or some equivalent, has come to be recognized. This 
field, in the tradition of some excellent earlier work antedating its emer- 
gence, has been primarily one of empirical research, offering opportuni- 
ties for fact-finding and for inductive generalization. By now, a rather 
considerable effort has been expended in these directions. It is natural 
to inquire what, as the result of fifteen years of such effort, we have added 
to our knowledge. 

The "price policy" field has been concerned in general with certain 
aspects of the manner in which business firms, singly and in groups, 
mobilize scarce resources to meet the effective demands for commodities. 
It naturally emphasizes description, explanation, and evaluation of the 
behavior of firms and industries in determining selling prices, outputs, 
and closely related matters, but it may appropriately encompass a good 
deal more. It has evidently taken its present form because of some shift 
in emphasis in investigating the affairs of business; as a new "field," 
it is essentially the result of a reorganization of one or more pre-existing 
fields, undertaken as a result of this altered emphasis. Since a part of the 
potential contribution of price policy research lies in its novel orientation, 
its antecedents may deserve some brief attention. 

Business organization and behavior have been studied intensively in 
this country since the early "merger movement," and even the earliest 
treatments were in general concerned with all of what we currently 
regard as the primary issues: (1) the structure, organization, and owner- 
ship of business; (2) the competitive behavior and price policies of enter- 
prise—including motives, strategy, and tactics; (3) the price, output, and 
associated results of this behavior; and (4) the public policy issues raised 
by such structure, behavior, and results. But the earlier work had a 



number of characteristics which were possible barriers to effective anal- 
ysis. It was excessively compartmentalized— there was a "natural mo- 
nopoly" utility field, a "trust problem" field, which encompassed industry 
which was not— but should be— "competitive," and special fields treating 
agricultural pricing, marketing institutions, and financial enterprise. 
Much of the work prior to 1930, moreover, featured simple description 
and superficial interpretation of financial organization, structural change, 
and competitive tactics, and in evaluating behavior emphasized the 
norms of law rather than those of economics. In it there was frequently 
a lack of close or extended analysis of price-output results, or of how 
observed market structure and competitive behavior affected the deter- 
mination of prices and outputs. There were, of course, some notable 
exceptions, where the tools and criteria of the available economic theory 
were diligently applied, or where a largely ad hoc analysis of industrial 
behavior moved directly to essential economic phenomena. But the short- 
comings noted in general reoccurred quite systematically. 

These limitations were especially apparent in the trust problem and 
marketing fields. They arose primarily from a general lack of raffort 
with the corresponding field of "economic theory," and this in turn 
stemmed from the "institutionalist" bias of writers, from their frequent 
lack of theoretical training, and from the inadequacy of contemporary 
price theory. It was thus that industrial concentration and collusion 
could be viewed as aberrations from a competitive norm, the theoretical 
validity and precise content of which was seldom examined, and that 
the real significance of these aberrations for the general material welfare, 
presumably registered through alteration of price-output results, could 
be left without real analytical evaluation. 

The emergence of the price policy field from these antecedents is 
traceable to several influences: (1) the experience of the Great Depres- 
sion, which engendered a more critical attitude toward the operation of 
business institutions and the character of competition; (2) the N.R.A. 
episode, which exposed a large number of economists to the more inti- 
mate details of pricing and competition in American business; and (3) 
the reformulation of price theory to make it more congruent with actual 
business behavior. Chamberlin's work seems to have been by all odds 
the most important in the last regard. It related the theory of pricing 
specifically to the institutional framework and practice of the real econ- 
omy—to concentration, product differentiation and its legal framework, 
collusive activities, trade practices, and barriers to entry; it predicted the 
probability of systematic and significant variations in competitive be- 
havior and price- and output-results in response to variations in this 


framework; and it thus posed many new questions for empirical research. 
In fact it provided a major "revelation" to many, suggesting in a broad 
stroke a general theoretical interpretation of the economic significance 
of the developments of business institutions since the beginning of the 
merger movement. 

Mrs. Robinson's work, although of a more formalistic character, was 
also influential; and such contributions as Zeuthen's Economic Warfare, 
though not widely read in this country, reveal a general tendency toward 
reformulation of price theory. The great formal contribution of Chamber- 
lin, however, was in recognizing adequately the possible economic 
implications, with respect to price, output, product, and cost, of the gov- 
erning institutional framework and of market structures, and in suggest- 
ing the necessity of an analysis employing several variables, several 
functional relationships, and several dimensions for the interpretation of 
actual price behavior. So interpreted, the doctrines of monopolistic and 
imperfect competition called for a new focus for empirical research in 
"institutional" fields. The relevance to ethical evaluation and public 
policy of this increased emphasis on patterns of price behavior (as op- 
posed to institutional emphasis per se) was in turn quickly recognized 
by the expansion of general equilibrium and aggregative analysis to 
point up the impact of quasi-monopolistic pricing on various dimensions 
of the total material welfare. 

The development of the price policy field then involved a certain 
reshaping of research emphasis. The study of matters of economic struc- 
ture, business organization and ownership, corporate finance, and the 
like has been given a separate though related status. Price policy research 
has tended to be focused mainly on certain aspects of behavior in all types 
of enterprise not under government price regulation— "competitive" busi- 
ness as well as "trusts," marketing as well as manufacturing firms, inter- 
national as well as intranational business arrangements. (The fields of 
regulated enterprise have been left largely compartmentalized, although 
this is not necessarily a good permanent arrangement.) The emphasis 
has been on the analysis of price-making and competitive behavior, its 
origins, and its results in output, price-cost relations, profits, selling costs, 
price flexibility, progressiveness in technique or product, and so forth. 
The terms and criteria of price analysis have been introduced into the 
empirical study of industry in pursuit of a precise knowledge of the 
economic consequences of institutional situations and practices, and. to 
elevate this study from the level of casual description to that of systematic 
generalization and explanation. 

The price policy field thus obviously transcends the narrower impli- 


cations of its label. It is not confined to a study of choices of alternatives 
by business managements in setting prices, outputs, products, and selling 
costs, or to an examination of processes of price calculation and of co-oper- 
ative activities with rivals. It embraces these things, but necessarily gives 
equal attention to the origins of observed behavior, to its results in terms 
of output and of price-cost and similar relationships, and to the evaluation 
of behavior and results from the standpoint of total welfare. In short, it 
is concerned in detail and on an empirical level with the range of prob- 
lems with which price theory, broadly construed, deals. It need not be 
limited, of course, to the formulations and concepts of any particular 
version of a priori theory. 1 

For purposes of this review, we will not view the price policy field 
as embracing the general treatment of public regulatory policy toward 
business competition and pricing. The latter is certainly a closely related 
field, although it appropriately partakes as much of political as of eco- 
nomic science. Even excluding it per se, however, we must recognize that 
regulatory experience is one of the primary sources of data for price 
policy study, and conversely that it is this study which can identify and 
analyze what is to be regulated, and may indicate what the economic 
effect of various policy measures will be. Thus any basic account of devel- 
opments in the price policy area must recognize the contribution of 
regulatory experience to research and analysis, and the effect of the latter 
on thought regarding public policy. 

Research within a "price policy" area thus defined has been proceeding 
rather intensively since the earlier i93o's. We therefore address ourselves 
to the main question: What has been accomplished since then? What 
do we know about business price making, its origins, its results, and their 
significance, that was not known in 1933? Since the price policy field as 
defined is primarily one of empirical research (as opposed to abstract 
theorizing per se), this question should be taken to refer primarily to 
inductive and empirical knowledge, or to knowledge stemming from and 
extensively supported by empirical data. We need not limit ourselves, 
however, to conclusive and established findings. We will therefore in- 
quire in turn into contributions in the form of: 

(1) Additions to scientific knowledge— in the form of empirical gen- 
eralizations, verification of hypotheses, or simply increments to infor- 
mation as yet not fully interpreted— concerning price making and 
competition, their origins, and their immediate results. 

1 Our view of the field as excluding a ■priori price theory proper is a provisional one, 
adopted in order to contain the present discussion within workable bounds. 


(2) New hypotheses, issues, and directions for investigation into these 

(3) Developments of method of research. 

(4) Developments of standards for the evaluation of price and re- 
lated results from the standpoint of their effects on total economic wel- 

(5) Suggestions pertinent to public policy. 

These headings should cover the main phases of work within the field 

Before we proceed to an appraisal of findings, however, an initial note 
of justifiable pessimism may be in order. Research in price policy has not 
yet found unity of direction; the field is still in the main a poorly charted 
area for exploration. Some forays have been made into its interior, but a 
large amount of effort has been expended at the borders in discussing the 
desirability of exploration, the equipment for the trip, the things to which 
the explorer might give attention, and the methodology of exploration in 
general. We are currently long on hypotheses and relatively short on dis- 
covery. Correspondingly the field has as yet no definitive work or works, 
nor has research as yet taken on such definite form that one can with 
impartiality construct a list of "important" (and thus of unimportant) 
contributions. Evaluations of particular items must be unusually subjec- 
tive. Succeeding mentions of individual works are thus definitely not 
viewed as "academy awards" of merit, but are made as expedient in chart- 
ing the main outlines of price policy research since 1933. 

I. Empirical Findings 

A field of empirical research is one in which students attempt to find 
out in fact what happens, in fact why it happens, and in fact what it leads 
to. The emphasis is on measured result, measured association, measured 
consequence. In approaching such research, however, there is no valid 
objection to referring to some a priori system of predicting behavior. In 
fact, such a system may be almost indispensable in suggesting directions 
for empirical work. 

The theories of monopolistic and imperfect competition, which pro- 
vided an initial orientation and stimulus for much price policy research, 
analyze pricing and production in three major steps. First, they point to 
certain significant aspects of the "structure" of markets— with emphasis 
on numbers of sellers and buyers, product differentiation, and ease of 
entry— which presumably influence competitive behavior and price re- 
sults. Second, they deduce the price-calculating and competitive or col- 


lusive behavior associated with various sub-categories of markets, as 
classified on the basis of these characteristics. They do this largely by 
attributing certain explicit sellers' demand curves, cost curves, and de- 
mand-selling cost relations to various market situations, and by deducing 
the profit-maximizing adjustments to these "determinants" of behavior. 
Third, they thereby predict the results attributable to various market 
structures or situations. 2 These predictions, in the absence of more precise 
data for "assumptions" than are ordinarily available, are almost entirely 
qualitative in content. 

Empirical price research has taken its approach in some part directly 
from the form of this theoretical analysis. One obvious opportunity has 
been to inspect various market structures in order to learn what theoreti- 
cal "types" exist in fact, and thus to provide price theory with a more 
relevant set of assumptions. Another opportunity is to make statistical 
measurements of demand curves, supply curves, and other presumed de- 
terminants of firm and industry behavior, and to use them to analyze 
actual price determination; and a third is to ascertain price and output 
results in various industries. All of these steps may implement the appli- 
cation of price theory or help verify its predictions. Not all inductive 
analysis of price policies, of course, has been limited to so narrow a for- 
mula. Market structures may be analyzed in an endeavor to establish a 
broader environmental base with which to link competitive behavior, and 
the analysis of this behavior may not be restricted to the ascertainment of 
demand and supply functions, but may deal in a freehand and experi- 
mental fashion with the price-calculating and competitive action of firms. 
Explicit attention may be given to matters often only implicitly subsumed 
in a friori analysis, such as the influence of public opinion and of govern- 
ment attitude on price policies, and the relationship of wage policies and 
price policies. Finally, price research may attempt to find on an empirical 
level demonstrable connections among market structure, competitive and 
price-calculating behavior, and price results. Work along this line might 
ultimately verify the system of abstract theory, or elaborate it, or replace it 
with something else. 


The narrower of the tasks mentioned have so far received much of the 
attention of students. A first major increment to our knowledge has been 
in the description and appraisal of American market structures, often 

2 We reserve for later consideration the accomplishment of general and aggregative 
analysis in relating immediate price-output results, predicted by particular equilibrium 
price theory or found in fact, to the total material welfare. 


with especial reference to the formulations of price theory. In line with 
the growing emphasis on the significance in pricing of oligopoly (fewness 
of sellers within industries) and oligopsony (fewness of buyers), princi- 
pal attention has been given to the number of sellers or buyers in various 
industries and to their proportionate control of output or sales, but em- 
phasis has also been put on product differentiation among rival sellers. A 
number of surveys dealing with or touching on concentration and other 
characteristics of market structures for broad samples of American indus- 
try have appeared, including the Thorp 3 and Wilcox 4 monographs, the 
Means 5 study of structure, the Hoffman monograph on food processing 
and marketing, 6 and the Smaller War Plants study of concentration. 7 For 
particular industries, more detailed analyses of market structure have 
been made, emphasizing geographical, technological, legal, and other 
characteristics of structure, as well as concentration and product differ- 
ence. The T.N.E.C. hearings and monographs, selected Federal Trade 
Commission material, and various other industry studies provide a fairly 
detailed appraisal of market structure in such industries as steel, alumi- 
num, cement, petroleum, automobiles, rubber tires, building materials, 
electric appliances, liquor, cotton textiles, bituminous coal, motion pic- 
tures, and others. Further, recent studies of cartels have brought together 
considerable information upon the structure of international markets for 
a number of commodities and manufactured goods. 

An obvious contribution of this descriptive work, 8 from the standpoint 

s W. L. Thorp and W. F. Crowder, The Structure of Industry, T.N.E.C. Monograph 
27 (Washington, 1940). 

4 Clair Wilcox, Competition and Mono-poly in American Industry, T.N.E.C. Mono- 
graph 21 (Washington, 1940). 

5 National Resources Committee, The Structure of the American Economy, Part I 
(Washington, 1939), especially Ch. 7, App. 6. 

A. C. Hoffman, Large Scale Organization in the Food Industries, T.N.E.C. Mono- 
graph 35 (Washington, 1940). 

7 Smaller War Plants Corporation, Economic Concentration and World War II, Sen. 
Doc. 206, 79th Cong., 2nd Sess. (Washington, 1946). 

8 By taking census commodity categories, singly or in related groups, as the principal 
basis for defining the industry, the writers of such studies have in most cases implicitly 
sought some working adaptation of a Marshallian-Chamberlinian definition of an indus- 
try—a group of sellers with identical or close substitute outputs having access for the bulk 
of their sales to a common group of buyers. Although empirical application of this con- 
cept necessarily involves some arbitrary disregard of measurable cross-elasticity of demand 
as among sellers in different industries, such cross-elasticities as must be neglected are in 
practice ordinarily not so important as to reduce greatly the analytical validity and useful- 
ness of the industry concept. In some cases, however, thoughtless handling of commodity 
classifications and neglect of geographical segmentation of the producers of related out- 
puts into local industries have resulted in inaccurate implementation of the industry 
concept, and in these cases the published results require careful reinterpretation. The 
following comments rest upon such a reinterpretation of findings wherever this has been 
necessary in order to retain an analytically valid approximation to the concept of an 


of price and output study, is that it has provided a fairly comprehensive 
measure of market structures in those dimensions which appear theoreti- 
cally to be most important in conditioning competitive behavior and price 
results. It shows how concentrated or unconcentrated various industries 
are and how differentiated their products are, and it supplies less com- 
plete information on relevant geographical and technical matters and 
conditions of entry. Referring directly to theoretical models, it suggests in 
what industries we have monopoly (one seller), monopsony (one buyer), 
oligopoly (a few sellers), monopolistic competition (a number of sellers 
with differentiated products), bilateral monopoly (one seller vs. one 
buyer), and bilateral oligopoly (a few sellers vs. a few buyers), and it 
furnishes a tentative basis for possibly desirable subdivisions and elabora- 
tions of the few simple categories of theory. 

The principal general indications of studies of American market struc- 
ture are (1) that concentration of output among relatively few sellers is 
the dominant pattern, (2) that fewness of buyers is common in producer 
goods markets, (3) that product differentiation is significant for practi- 
cally all consumer goods and a number of producer goods, (4) that there 
are potentially many significant sub-varieties of "fewness" and concentra- 
tion which would logically fall within the bounds of the oligopoly (or 
oligopsony) category, 9 and (5) that there are additional market charac- 
teristics, such as the durability of the output, the geographical pattern, 
the degree of imperfection in market organization, and several others 
upon the basis of which markets might be meaningfully distinguished. 
Such findings suggest that pure competition, many-small-seller monopo- 
listic competition, and single-firm monopoly are in practice rather special 
cases, and that oligopoly, as the general case, may require elaboration and 
subdivision. 10 

These findings may thus suggest the elaboration of old or development 
of new theoretical models appropriate to particular cases, by feeding more 
precise or elaborate "market structure assumptions'' into the deductive 

For example, very few sellers with equal shares (three, each with a third of the 
market); very few sellers with distinctly unequal shares (five, with the largest seller con- 
trolling two-thirds of the market); moderately few sellers (ten or fifteen) with various 
patterns of concentration; "quite a few" sellers (twenty to forty or more) with various 
patterns of concentration; the concentrated core with the competitive fringe (four sellers 
control 80 per cent of the market, and thirty small sellers divide the remainder); and 
so forth. 

10 These findings do not ordinarily reveal in which oligopolistic industries the sellers 
have explicit or tacit collusion on price or other matters, or the degree or effectiveness 
of the collusion. It is thus expedient, as well as theoretically scrupulous, to define oligop- 
oly as occurring wherever there are a few firms (i.e., separate ownership interests) selling 
within an industry, regardless of the extent of collusion among them. The emergence 
of collusion is best viewed as a phenomenon of competitive behavior, apart from market 


mill. They may also tell us which models from developed price theory are 
applicable to given actual market situations, and they may allow us to 
predict (so far as the models are reliable) the general quality of price 
results forthcoming from these markets. 

There is in fact a variety of applied price theory which involves little 
more empirical work than that already mentioned. The authors of a num- 
ber of interesting studies have investigated particular market structures 
mainly in search of the appropriate assumptions for theory. Having found 
these, they have proceeded to construct the implied demand and supply 
situations and by deduction to predict (occasionally with hypothetical 
cost and price data) the price results attributable to the markets studied. 11 
This is of course empirical research that is not very empirical, in that it 
leaves the substantial matters of competitive behavior and price results 
to hypothesis. It is an improvement over abstract theory which leaves its 
assumptions to guesswork or to excessively casual observation, but it is 
not a very reliable means of finding out what happens in the way of com- 
petition and pricing. The dominance of oligopolistic situations in actual 
markets implies in effect that reliable a priori predictions may not be 
available. Even if oligopolistic indeterminacy is disposed of, moreover, 
the interpretation or prediction of monopolistic prices which are charged 
through time and are dependent on a half-dozen variables, each of which 
is subject to uncertainty, evidently requires a much more complex formal 
theory than has yet been developed. Despite their merit as elaborations 
and adaptations of conventional theory, therefore, the last-mentioned 
studies have added to our empirical knowledge of price determination 
mainly by acquainting us with the setting of pricing action. Other studies 
which have stopped short of forays into hypothetical demand and cost 
curve construction have been almost as informative as those which have 
pushed on. Studies of market structures as a group have provided a 
knowledge useful in applying abstract price analysis to the real economy 
or as a basic starting point for further empirical investigation. 


Beyond the basic market structures, the hypothetical determinants of 
pricing action are industry and individual-seller demand curves, cost 
curves, and other relationships between interdependent variables. A 
natural focus for price research has thus been on the quantitative deter- 
mination of these relationships from statistics. Accurately determined, 

11 See, for example, W. H. Nicholls, A Theoretical Analysis of Imperfect Competition 
with Special Application to the Agricultural Industries (Ames, 1941) (a study contain- 
ing some very interesting treatments of bilateral oligopoly); also H. B. Meek, "A Theory 
of Hotel Room Rates," Hotel Administration, June 1938, IX, 


statistical demand, cost, and other functions might enable us to check the 
validity of abstract theoretical predictions (so far as these have objective 
content) and, if the theory is verified, to make objective and quantitative 
predictions of behavior. Several statistical studies of demand and cost 
with direct bearing on price policy matters have appeared, and their gen- 
eral content may deserve note. 

Studies of demand might inquire into price-sales relationships either 
for individual sellers or for industries or related groups of sellers. In view 
of the theoretical and statistical problems involved, it is not surprising 
that there are practically no empirical studies of demand curves or similar 
relationships for individual sellers. 12 The quaesitum if conventional 
theory is to be implemented is the ex ante or anticipated demand curve 
for the seller's output. This might be sought either by attempting to find 
an objective ex post demand relation, which might be hopefully taken as 
a fair approximation to the seller's anticipated curve, or by inquiring into 
and attempting to quantify the seller's subjective impression of his future 
demand. Neither approach is very promising. In oligopolistic industries 
where there is no express or tacit collusion on price, the individual-seller 
demand curve is on realistic assumptions indeterminate and could not be 
uniquely ascertained by any means. In collusive oligopoly 13 or in non- 
oligopolistic industries the curves are hypothetically determinate, but the 
data from which such functions might be determined ex post are ordi- 
narily unavailable or inaccessible, except where, in collusive oligopoly, 
the seller's demand curve might be determined as some share of the ex 
post industry demand, if the latter were known. Even should an ex post 
seller's demand curve be established, moreover, it might often be a rather 
poor approximation to its ex ante counterpart (if any) and further fail to 
reflect the fact that a range of alternative estimates may replace single- 
valued estimates in an uncertain world. Direct inquiry into sellers' sub- 
jective impressions of their future demand curves seems to this writer 
unlikely to yield reliable quantitative results useful in explaining price. 14 
It may thus be legitimate to conclude that the individual-seller demand 
curves which hypothetically influence pricing will not ordinarily be reli- 

12 But see the interesting exploratory work of R. M. Whitman, "Demand Functions 
for Merchandise at Retail," Studies in Mathematical Economics and Econometrics, 
Henry Schultz Memorial Volume, Lange et al., eds. (Chicago, 1942), pp. 208-221. 

13 Collusive oligopoly here refers to all oligopolistic industries with agreements, prac- 
tices, or formal or informal conventions whereby the several sellers obtain effectively 
concurrent and non-competitive action in setting and changing price, and possibly also 
in other matters. It thus embraces partial or imperfect collusion as well as full theoretical 
cartelization, and subsumes (in the view of this writer) most real oligopoly cases. "Fully 
recognized interdependence" ordinarily gives birth to collusion in this sense. 

14 Some of the reasons for this are discussed in the following section, pp. 154-155. 


ably found or approximated on a quantitative level, unless as a collusive 
oligopolist's "share" of a known industry demand. Applied price analysis 
will ordinarily have to rely, if such reliance is justified, upon the statisti- 
cally found ex post industry demand curve and upon an analysis of com- 
petitive and co-operative adjustments of rival sellers to this demand. 

There are, of course, a number of studies of "industry" demands for 
products, measuring the combined demand for the outputs of several rival 
sellers on the implicit, and ordinarily supportable, assumption that they 
change their prices concurrently and by similar amounts. Leading exam- 
ples of these in the industrial field are the General Motors study of auto- 
mobile demand, 15 and the Yntema study of steel demand, 16 both prepared 
at the behest of leading producers in concentrated industries as they made 
ready to defend themselves against claims that they could reduce prices. 
A common difficulty with these studies has been that the basic quantity 
and other data occur in such form (often in annual aggregates) and be- 
have and are interrelated in such fashion that they are not amenable to 
very meaningful treatment by the partial and multiple correlation tech- 
niques ordinarily employed. As a result no especial meaning can be at- 
tributed to most statistical industry demand curves which have been or 
could be forced out of available statistical data. The main positive contri- 
bution of such studies, in fact, has come from largely qualitative com- 
ment on and interpretation of particular industry demands. Thus the 
General Motors study illuminates very well the complicated behavior of 
the demand over time for an expensive durable good where there is a 
used-product market. The Yntema study develops a convincing case for 
the inelasticity of steel demand largely by qualitative analysis of the char- 
acter of its uses. Experience in this field so far has suggested (i) that 
statistical measurements of industry demand curves will ordinarily be so 
unreliable (because of intercorrelations of variables, large probable errors, 
etc.) that they cannot be regarded as quantitative information usable in 
further analysis; (2) that an indirect and often qualitative approach to 
the evaluation of industry demands will often have to suffice; (3) that 
preponderant attention to net price-quantity relations, with comparative 
neglect of, or subordination of emphasis on, more complex relations, may 
often unnecessarily restrict applied analysis; 17 and (4) that the interpre- 

10 General Motors Corporation, The Dynamics of Automobile Demand (New York, 

16 United States Steel Corporation, A Statistical Analysis of the Demand, for Steel, 
1919-1938, T.N.E.C. Papers (New York, 1939). See also other T.N.E.C. Papers of 
U.S. Steel. 

17 The complex multi-variate relations, together with partial relations in addition to 
that between price and quantity, have frequently been calculated in statistical demand 


tation of price policies should take explicit account of the uncertainties 
and complexities noted. Moreover, any statistical industry demand curve, 
no matter how reliable, will almost inevitably be determined ex 'post, and 
its correspondence to the industry demand anticipated by sellers may not 
be especially close. Finally, even possession of an industry demand curve 
which is regarded as a reliable approximation to the mean expectation 
upon which sellers acted would be only a beginning in the process of 
verifying the predictions of abstract analysis. For this, the competitive 
and co-operative adjustments of sellers to this demand must be analyzed; 
and on this level it may often be extremely difficult, with manifold un- 
certainty and with a multi-dimensional price-policy problem, to fill in the 
hypothetically determining relationships suggested by simplified theory. 
Statistical studies of costs have also been made, particularly of the 
short-run net relation of production cost to output for the firms, as em- 
phasized in conventional price theory. Much of the progress along this 
line has been made by Joel Dean, who in a number of monographs 18 has 
studied the short-run cost functions of firms in various industries, and 
has added considerably to factual knowledge of costs as well as to the 
refinement of applied statistical analysis. Statistical difficulties are en- 
countered with costs as with demand— they loom large when analysis 
must depend upon a few annual observations, as it did in the Yntema 
study 19 of steel costs— but on the whole they are fewer, and the results 
obtained appear to be more reliable. One striking finding of such cost 
studies has been the apparent linearity, over wide ranges of output, of the 
short-run total production cost functions of observed firms, implying a 
constant short-run average variable and marginal cost except at extremely 
small or large outputs. This has in turn given rise to attacks on the valid- 
ity of the finding, with some emphasis on the predisposition of statisti- 
cians to use linear regressions in correlation analysis, and on the tendency 
of accounting classifications and allocations of costs to bias the cost-output 
relation in the direction of linearity. Although there has been some merit 
in these criticisms as applied to certain studies, the better cost-curve 
analyses, like Dean's, have defended their findings rather well. Although 
the sample of firms studied is still extremely small, it would appear that 
for a large group of firms with multiple-unit plant equipment the propor- 

studies, but have often been put aside and neglected as principal attention turned to the 
demand curve and its price elasticity. 

18 See Joel Dean, Statistical Determination of Costs, with Special Reference to Mar- 
ginal Costs (Chicago, 1939), and a number of subsequent monographs developing 
statistical cost curves for different types of firms. 

19 United States Steel Corporation, Steel Prices, Volume, and Costs, T.N.E.C. Papers 
(New York, 1939). 


tion of variable to fixed factors may be held relatively constant over a 
wide range of output, and that as a result short-run average variable and 
marginal costs may be approximately constant over a similar range. This 
finding, if further established, is relevant to price analysis. However, 
since the bulk of the firms in question operate in concentrated industries 
where the tendency to extend output is fully checked either by the de- 
cline of price with increasing output or by increasing marginal selling 
costs as more is sold at given prices, the finding is hardly revolutionary in 
its implications. The resistance offered to it suggests that some critics 
have been unduly enamored of the symmetrical U-shaped cost curves of 
conventional a priori analysis. As yet, the sample of firms and industries 
studied is very small. The data uncovered are interesting fer se, have 
been useful in giving us a more accurate idea of the actual shapes of the 
functions emphasized in abstract theory, and have cast some added light 
on the phenomenon of price rigidity. They have not been put to more 
than casual use in the analysis of the determination and behavior of spe- 
cific prices, and their potential utility in this regard remains to be ex- 

The "long-run" relation of cost to output, or to scale of plant or firm, 
has been studied in rough qualitative fashion in several industries, and 
with thoroughgoing statistical analysis in at least one instance. 20 More 
comprehensive studies made of this sort of relation— for example, the 
Federal Trade Commission study of size and efficiency— are fragmentary, 
based on unrefined data, and substantially worthless. 21 Engineering esti- 
mates of the cost-scale relationship, most of which are made for internal 
control purposes and remain unpublished, may constitute the best source 
of information on this subject. 

Although beginnings have been made in the statistical ascertainment 
of cost and its relation to other variables, our knowledge so far is frag- 
mentary and incomplete. A price committee of the National Bureau of 
Economic Research underlined this conclusion in an extensive survey 22 
of our inadequate empirical knowledge of costs, and pointed to desir- 
able directions for additional research effort in this field. A much more 

20 Joel Dean and R. W. James, "The Long-Run Behavior of Costs in a Chain of Shoe 
Stores— A Statistical Analysis," Journal of Business, University of Chicago, April 1942, 

21 Federal Trade Commission, Relative Efficiency of Large, Medium-Sized, and Small 
Business, T.N.E.C. Monograph 13 (Washington, 1940). Some of the principal short- 
comings of this work involve the use of insufficiently rectified accounting costs as basic 
data, and the non-recognition of differences other than those in scale of plant as the 
probable causes of the noted differences in average costs among plants of various sizes. 

22 Conference on Price Research, National Bureau of Economic Research, Cost Be- 
havior and Price Policy (New York, 1941). 


complete knowledge for many industries of cost-output relations, both 
short- and long-run, of the relation of cost to other variables, and of the 
composition of costs would indeed be useful. Substantially complete 
knowledge of such matters, however, would not lead at all easily or di- 
rectly to an explanation of price behavior in terms of statistically found 
price-determining functions. Granted that ex fost cost functions may 
represent workable approximation to the theoretically determining ex 
ante functions, comprehensive statistical testing of conventional theory 
requires also reliable industry and individual-seller demand curves, and 
as noted these have not been, and very well may not be, filled in. Unless 
they are, the processes of existing theoretical price analysis could cer- 
tainly not be reproduced on an empirical level. Even if this barrier could 
be crossed, however, verification of simplified two-dimensional theory 
based on the assumption of given data should hardly be expected. Any 
applicable theory must be of a multi-variate, multi-dimensional character, 
including numerous functional relationships in addition to those between 
cost and output and between price and output, and quantitative verifica- 
tion of such elaborated models in an uncertain world may not, as we will 
note in a succeeding section, be at all feasible. (In developing and inter- 
preting their findings, nevertheless, statisticians might be less influenced 
by the two-dimensional formulations of textbook theory, and might em- 
phasize more the multi-variate relations with which they always neces- 
sarily deal in processing basic data.) Statistical cost studies promise to be 
most useful, for the time being at least, as general data in a type of ap- 
plied analysis which from the standpoint of simplified theory appears to 
be of a catch-as-catch-can variety. 


A broader undertaking is the general empirical study of pricing and 
competition in the individual industry. Industry studies were made, of 
course, before 1933 and a number of them, including various Federal 
Trade Commission studies, were of very good quality. They were gener- 
ally marked, however, by an extensive preoccupation with matters of 
finance and organization, by an interpretation of competition largely in 
legal terms, and by an attention to little else in the way of price results 
than profits. The later industry studies, in the "price policy" tradition, 
tend to accept the terms of Chamberlinian and post-Chamberlinian price 
theory at least as an orientation for investigation and sometimes as a spe- 
cific formula for verification. Thus the "full-dress" industry study of re- 
cent years pays attention to the character and historical origins of market 
structure— including concentration, product differentiation, entry, tech- 


nological and legal conditions, and geographical relationships— to the 
resultant conditions of demand and supply for the industry and for firms, 
to the character of price calculation and of rivalrous and collusive be- 
havior, and to price results including profits, efficiency, selling costs, pro- 
gressiveness, price rigidity, price discrimination, and so forth. And it 
often also examines at some length the interconnections of structure, be- 
havior, and results— in effect, it moves in an empirical case study through 
most of the range of interests of the price policy field. 

This acceptance of the terms of theoretical price analysis is a blessing 
and can be a curse. It is a blessing in that it provides a systematic orienta- 
tion for the empirical study of industry behavior. It provides initially a 
reasonably pertinent set of precise questions, the answers to which will 
provide data for an evaluation of how, relative to its opportunities, the 
industry utilizes resources for production. It can be a curse if the investi- 
gator becomes preoccupied with an unduly narrow range of questions, as 
propounded by a simplified theory, or loses sight of the fundamental im- 
portance of the institutional conditions and arrangements which condi- 
tion behavior, concentrating unduly on a maze of intersecting plane 
geometry which after all can only reflect the force of these fundamental 

It is a tribute to the good sense of workers in this field that in all but a 
few studies they have been able to assimilate the essential import of 
modern price analysis without letting it obscure the significance of insti- 
tutions or the fact that economics really deals with human beings and 
physical things rather than with mathematical symbols. The coverage of 
industries in empirical studies to date of course gives us only a small 
sample of the possible total. Book-length studies are available for such 
industries in the United States as aluminum, 23 steel, 24 newsprint paper, 25 
petroleum, 26 butter and margarine, 27 motion pictures, 28 cigarettes, 29 

23 D. H. Wallace, Market Control in the Aluminum Industry (Cambridge, Mass., 
J 937)- See also, for more recent data, N. Engle et al., Aluminum, An Industrial Market- 
ing Appraisal (Seattle, 1944); and C. F. Muller, Light Metals Monopoly (New York, 

24 C. R. Daugherty, M. G. de Chazeau, and S. S. Stratton, The Economics of the 
Iron and Steel Industry, 2 vols. (New York, 1937). 

25 J. A. Guthrie, The News-print Paper Industry (Cambridge, Mass., 1941); idem, 
"Price Regulation in the Paper Industry," Quarterly Journal of Economics, February 
1946, LX, pp. 194-218. 

26 J. S. Bain, The Economics of the Pacific Coast Petroleum Industry, 3 vols. (Berke- 
ley, 1944-47). 

27 W. R. Pabst, Butter and Oleomargarine (New York, 1937). 

2S M. D. Huettig, Economic Control of the Motion Picture Industry (Philadelphia, 

29 R. Cox, Competition in the American Tobacco Industry, 1911-1932 (New York, 


oranges, 30 automobiles, 31 and cotton textiles, 32 and briefer article-length 
surveys for such industries (or parts thereof) as electric lamps, 33 cran- 
berries, 34 potash, 35 bakeries, 36 rubber tires, 37 electric appliances, 38 marga- 
rine, 39 barbering, 40 and farm machinery. 41 These are supplemented by a 
number of studies of international cartel organization and operation, 42 
as well as by more fragmentary contributions of various sorts. In addition, 
several volumes have surveyed the character of competitive behavior and 
its results for a number of industries or firms. These include studies by 
Burns, 43 Hamilton, 44 and Nourse and Drury, 45 all of which, though nec- 
essarily cursory in empirical analysis of particular industries, have been 
influential in shaping thought and stimulating effort within the field. 
Finally, there has appeared a considerable amount of material which 
bears on competitive behavior and pricing but is in rather unorganized 
form, including T.N.E.C. industry materials not previously cited— in 
particular the hearings on the steel, petroleum, liquor, construction and 
other industries— and various publications of the war agencies during the 
past several years. 

The number of items in this representative sample suggests the im- 

30 D. A. Revzan, The Wholesale Price Structure for Oranges, with Special Reference 
to the Chicago Auction Market (Chicago, 1944). 

31 Federal Trade Commission, Report on the Motor Vehicle Industry (Washington, 


32 S. J. Kennedy, Profits and Losses in Textiles (New York, 1936). 

33 A. R. Bright and W. R. Maclaurin, "Economic Factors Influencing the Development 
and Introduction of the Fluorescent Lamp," Journal of Political Economy, October 1943, 
LI, pp. 429-450. 

34 C. D. Hyson and F. H. Sanderson, "Monopolistic Discrimination in the Cranberry 
Industry," Quarterly Journal of Economics, May 1945, LIX, pp. 330-369. 

35 S. P. Hayes, "Potash Prices and Competition," Quarterly Journal of Economics, 
November 1942, LVII, pp. 31-68. 

36 L. G. Reynolds, "The Canadian Baking Industry; A Study of an Imperfect Market," 
Quarterly Journal of Economics, August 1938, LIII, pp. 659-678. 

37 Idem, "Competition in the Rubber Tire Industry," American Economic Review, 
September 1938, XXVIII, pp. 459-468. 

38 W. G. Keim and J. M. Blair, "The Electrical Equipment Industries," Chap. 4 
(pp. 109-164) of Price Behavior and Business Policy, T.N.E.C. Monograph 1 (Wash- 
ington, 1940). 

39 W. H. Nicholls, "Some Economic Aspects of the Margarine Industry," Journal of 
Political Economy, June 1946, LIV, pp. 221-242. 

40 W. F. Brown and R. Casady, "Guild Pricing in the Service Trades," Quarterly 
Journal of Economics, February 1947, LXI, pp. 311-338. 

41 J. T. Dunlop and E. M. Martin, "The International Harvester Company," Part II 
(pp. 63-137) of Industrial Wage Rates, Labor Costs, and Price Policies, T.N.E.C. 
Monograph 5 (Washington, 1940). 

42 See especially G. W. Stocking and M. W. Watkins, Cartels in Action (New York, 

43 A. R. Burns, The Decline of Competition (New York, 1936). 

44 W. Hamilton et al., Price and Price Policies (New York, 1938). 

45 E. G. Nourse and H. B. Drury, Industrial Price Policies and Economic Progress 
(Washington, 1938). 


practicality of any item-by-item evaluation of industry studies. Such 
studies as a group, however, allowing for individual differences in qual- 
ity, length, and scope, have resulted in a substantial accretion to our em- 
pirical knowledge of competition and pricing and of their origins and 
results. In a number of specific cases we can observe the significant char- 
acteristics of market structure, their effects on and reflections in condi- 
tions of demand and supply, the actual character of competitive and col- 
lusive action, and the statistical and qualitative measurements of the 
results which emerge. And we can appraise the apparent causal inter- 
connections between structure, competitive behavior, and price results. 

What is the net contribution to our general knowledge of business 
behavior which emerges from the industry studies now on hand? The 
verification and implementation of a 'priori price theory in its usual sim- 
plified form is not a primary nor a uniformly successful accomplishment. 
To be sure, it has been possible, on a strictly qualitative level, to support 
some of the broader hypotheses which theorists have advanced concern- 
ing the association of market structure with price making and its results. 
Distinct differences appear between price behavior in industries with 
atomistic structure and that in industries with a high concentration of 
sales in the hands of a few firms; the extent to which entry is blockaded 
seems actually important from the standpoint of profits; product differen- 
tiation and product variability are associated systematically with the be- 
havior of selling and production costs; industries with few buyers have 
had price results different from those with many buyers; and so forth. 
But it has been ordinarily impossible to establish on a quantitative level 
the actual relationship of marginal costs to marginal receipts, or otherwise 
to implement the precise analytical constructions of a priori analysis. It 
has been possible to fill in tentatively certain blanks where a priori theory 
makes no unique predictions, especially in the ubiquitous category of 
oligopolistic industries. The bogey of potentially indeterminate behavior 
in this category has been at least in part removed by an identification of 
specific patterns of collusive and quasi-collusive behavior so common in 
such industries, by examining the content and workability of such collu- 
sion, and by pointing to reasonably systematic time patterns of price-out- 
put behavior as emerging from such industries. But the main accomplish- 
ments of empirical industry study are of more far-reaching character. 

The first of these is in exploring in considerable detail the fundamen- 
tal institutional, technological, and geographical conditions which lie 
behind "demand and supply" and affect enterprise behavior. Conven- 
tional a priori theory deals with the conditions of demand and cost for 
firms and industries largely as independent objective determinants of 


behavior. It goes behind the veil of demand and supply to the more fun- 
damental determinants of behavior only in cursory fashion, referring 
briefly to such matters as the numbers of sellers and buyers, the degree of 
product differentiation, and the tendency to diminishing returns within 
the firm. Extended empirical research has pierced this veil much more 
thoroughly and with more careful observation, with the result that the 
general tendencies of price behavior are given a detailed explanation in 
terms not simply of demand and supply curves, but of the basic institu- 
tional, technological, and related conditions which shape and shift these 
determining functions. The optimum result might be characterized as a 
wedding of the theoretical and institutional approaches to economic 
behavior, or perhaps better as a theoretical interpretation of economic 
history. Thus in the study of the American aluminum industry, 46 the 
manner in which a monopolistic firm with certain costs adjusts its output 
to a given complex of demands is not the sole subject for investigation, 
and research is not confined to the statistical measurement of cost and 
demand functions. Attention is given in equal measure to the manner in 
which patent control, resource ownership, and international agreement 
combined to blockade entry and preserve monopoly; to the character and 
effect on demand of the substitute competition offered by copper and 
aluminum scrap; to the time trend of demand and its effects; and to the 
specific technological conditions which affect the relation of cost to scale 
and to integration. With the analysis placed on this foundation, the suc- 
ceeding interpretation of behavior can trace the effect of technique, law, 
institutional device, and historical accident through demand and supply 
and to observed price-output results. The resulting accomplishment, 
noted in a number of industry studies, is to preserve the utility of techni- 
cal analysis while avoiding the essential oversimplification and superfici- 
ality of its assumptions. 

A related contribution of empirical industry study is to reveal the com- 
plexity of competitive behavior and of its determinants in actual situa- 
tions, and thus to suggest reasons for wide divergences in the observed 
behavior of industries which would seem to fall within a single category 
in a "priori theory. This contribution is significant because such theory 
overlooks strategic characteristics of market structure in its descriptive 
assumptions and recognizes too few variables and relationships to allow it 
to make certain very relevant distinctions among cases, and also because 
the usual static equilibrium formulation passes over essential process and 
sequence phenomena in price, cost, and output behavior and in the 
dynamics of product and technique. Thus for example in the analysis of 

w D. H. Wallace, of. cit., especially Parts II and III. 


the automobile industry, 47 we find that very essential aspects of market 
structure and of competitive behavior have been touched on rather lightly 
in the theories of differentiated oligopoly. The problem of the determina- 
tion of price, output, product, and selling cost is not simply one of oli- 
gopolistic interdependence at a price nexus, but is dominated throughout 
by the extreme durability of the product, its high unit cost and price, and 
its susceptibility to style change. The emphasis in analysis thus turns 
heavily on the pattern of shift through time of the demand for the new 
product in response to initial exploitation of a latent demand, to partial 
saturation of the demand, and to replacement cycles. It turns also on the 
manner in which the behavior of this demand can be influenced by 
sellers through product variation, introduction of instalment selling, 
organization and control of a used-product market, and integration or 
control of retail sales outlets. The essence of price policies in the automo- 
bile industry is found in the concurrent and the competitive adjustments 
of sellers to the shifting potential demand situation through time. A free- 
hand analysis of these policies reveals the emergence in the 1930's, by a 
process of progressive imitation, of a fairly regular pattern of competitive 
and of concurrent behavior possibly peculiar to a concentrated durable 
consumer-good oligopoly in maturity. Comparative analysis of oligopolis- 
tic behavior in such industries as those producing cigarettes, soap, rubber 
tires, and petroleum products reveals equivalent complexities in the set- 
ting and character of price policies and re-emphasizes the necessity of 
interpreting behavior through time. It also suggests that the oligopoly 
category may be logically divisible into several significant parts on the 
basis of differences with respect to conditions of entry, number of buyers, 
differentiation, variability, and durability of product, trend of demand 
through time, and other market characteristics. It is of course true that if 
a priori theory is overly general, theorizing based on empirical analysis 
can as easily become overly specific. 

A third contribution of empirical industry study is of course in finding 
and measuring the significant results of competitive behavior in various 
industries. The ratio of profits to investment, of selling to production 
cost, of attained to optimum scale or utilization, and related results can 
be ascertained directly, and thus the strictly qualitative predictions of a 
friori theory can be replaced with a quantitative knowledge of the out- 
come of various patterns of competition within various market structures. 

47 See General Motors Corporation, on. cit.; W. Hamilton et al., op. cit., Sec. II (by 
M. Adams); Federal Trade Commission, Report on the Motor Vehicle Industry; H. B. 
Vanderblue, "Pricing Policies in the Automobile Industry," Harvard Business Review, 
Summer 1939, XVIII, pp. 385-401. 


A considerable accumulation of such factual knowledge can be of the 
utmost importance to economics. It is much easier to measure such re- 
sults, however, and to show their historical association with given situa- 
tions, than to demonstrate a necessary and causative relation with chosen 

An implication of the findings described is that it would be desirable 
in a 'priori theory to elaborate the basic assumptions by reaching further 
and in greater detail behind the veil of demand and supply into the tech- 
nological, geographical, and institutional context of business action— to 
take account of the influence of more characteristics of market structure 
than has been customary. A related implication is that the formal appara- 
tus of price theory should be further accommodated to the simultaneous 
treatment of numerous variables and their interaction through time. We 
will refer to this matter in the following section. 

There appear to be definite limitations, however, on the progress of 
work within the field of industry studies. It is only in the extended 
studies— like the Wallace aluminum study, which clearly established a 
landmark in approach, method, and quality early in the development of 
this field— that a detailed measurement and evaluation of results, or any 
reasonably conclusive analysis of how one thing leads to another, can be 
undertaken. Yet each such study is time-consuming in the extreme, and 
the specific results of each become outdated rapidly if the industry in 
question is in the process of dynamic change. Shorter studies are perhaps 
better adapted to the time limitations of research workers and to the 
tempo of economic change; but in such studies the view of behavior is 
often excessively synoptic, and frequently little more can be done than 
to establish the general correspondence of a particular industry with some 
theoretical model, to carry through a rough and usually unsatisfactory 
statistical analysis of conditions of industry demand and cost, and to 
attempt a necessarily sketchy mensuration of significant results. Even a 
large amassing of studies of the latter scope will hardly provide the basis 
for any very extended scientific generalization. 

Another difficulty concerns the availability of essential data. The great 
bulk of work in this field is perforce undertaken from the standpoint of 
the outsider, is based on published or other non-confidential materials, 
and proceeds by inferring as much as possible from this sort of data. The 
reluctance of businessmen to confide to economists their methods of price 
calculation and the character of their associations with rival firms— an 
attitude strongly encouraged in the United States by the anti-trust laws- 
has been a serious barrier to close investigation of price policy as seen by 
the price-maker. This attitude has very frequently been an obstacle to 


any conclusive empirical analysis of the processes of price and output 

A final limitation is evident if an optimistic view of potential accom- 
plishment has been taken initially. It seems improbable that work in this 
field will ever approach the level of closely approximate quantitative ex- 
planation and prediction of prices, outputs, and associated magnitudes. 
The important determinants of any one quantity are nearly always 
several, and although each determinant may have a potentially objective 
magnitude, it is generally subject to manifest uncertainty and to subjec- 
tive interpretation or estimate by decision-makers. The potential precision 
of explanation suggested by a friori price theory operating on the assump- 
tion of given data will thus probably never be approached. Empirical 
explanation must, moreover, early depart from the mold of static analysis 
of a single homogeneous time period and attempt to explain what hap- 
pens in a situation evolving through a series of interrelated time periods, 
where the "given" does not remain given and where process is as impor- 
tant as equilibrium tendency. The accomplishment of empirical research 
into business pricing is thus likely to take the form of a qualitative inter- 
pretation of a historical process, and correspondingly of qualitative pre- 
diction. Conventional price theory will be useful for the interpretation 
of tendencies implicit in market situations, but perhaps not as a frame- 
work for detailed quantitative analyses. 


Not all of the significant empirical findings are contained in intensive 
studies of industries or firms which trace behavior through from its 
origins to its final consequences. The Burns study, for example, which 
early emphasized the unity between institutional fact and the theory of 
monopolistic competition, centered largely on a systematic view of 
practices of competition and collusion, and the Hamilton collection is 
a freehand description of business behavior in a number of selected cases. 
Of an essentially similar character is the Borden study of advertising, 
a broad but somewhat cursory survey of the probable effects of advertising 
on demand, cost, price, and so forth for a number of consumer goods. 18 
Yet each was in its way penetrating and suggested a broadening and 
reorientation of inquiry into pricing and competition. Much essentially 
fragmentary and unorganized information on competitive behavior and 
tactics which has become available in the last decade and a half is also 
very suggestive of the character and complexities of competition. The 
digested results of the N.R.A. experience of course provided a mine of 

4S N. H. Borden, The Economic Effects of Advertising (Chicago, 1942). 


fact which has been very enlightening to economists and which has 
been drawn upon heavily in later research. Much the same is true of the 
findings in various actions of the Federal Trade Commission, and also 
of its interpretation of the effect on pricing and competition of various 
regulatory measures. 49 The large number of anti-trust cases investigated 
or tried after the inception of the Arnold regime in the Anti-Trust 
Division of the Justice Department, and the connected writings of mem- 
bers of that regime, 50 emphasized the importance of collusive activities 
as a part of the price-determining process, and strongly suggested the 
necessity of accommodating "technical" analysis to admit this sort of 
fact. The same general suggestions are implicit in the wealth of recent 
materials on international cartel activities. It is time that theorists for 
sook their cautious adherence to the improbable assumption that inter- 
dependent oligopolists remain "independent" and rely on a guessing 
game as the road to satisfactory industry prices. 

The experience of regulatory agencies during the war period— espe- 
cially that of the O.P.A.— has of course provided a potentially valuable 
mine of information for future price policy study. Most of the published 
literature bearing on this experience so far, however, has been of a 
specialized character and is primarily concerned with an atypical regu- 
latory problem. In the main it simply adds to the body of source material 
waiting to be tapped. Its considerable potentialities are emphasized by 
some retrospective surveys which have appeared recently. 51 

Any account of empirical price research must mention the extensive 
literature on commodity price rigidity. Originating with the Means 
pamphlet, 52 the controversy on price inflexibility was carried on in the 
journals and elsewhere for about a decade. As in the case of similar 
controversies, many of the arguments which were raised and finally 
settled or forgotten now seem relatively unimportant, and in retrospect 
it appears that the fashionable attention to this one aspect of price be- 
havior resulted in some imbalance in emphasis within the field of price 
study. The principal issues included : ( 1 ) how inflexible industrial prices 

46 See, for example, Report of the Federal Trade Commission on Resale Price Mainte- 
nance (Washington, 1945) which casts considerable light on the character of competi- 
tive and co-operative activity in the distributive trades, and on the effect of state fair trade 
laws upon prices and margins. See also E. T. Grether, Price Control Under Fair Trade 
Legislation (New York, 1939). 

50 See especially Thurman Arnold, Bottlenecks of Business (New York, 1940). 

51 See, e.g., J. K. Galbraith, "Reflections on Price Control," Quarterly Journal of Eco- 
nomics, August 1946, LX, pp. 475-489; also Richard B. Heflebower, "Content and 
Research Uses of Price Control and Rationing Records," American Economic Review, 
May 1947, XXXVII, pp. 651-666. 

62 G. C. Means, Industrial Prices and Their Relative Inflexibility, Sen. Doc. 13, 74th 
Cong., 1st Sess. (Washington, 1933). 


arc, cyclically and secularly, and whether they have become more in- 
flexible with the "decline of competition"; (2) why they are inflexible, 
in terms of the price-calculating operations and competitive relations of 
firms, and in terms of underlying market structures; and (3) what is the 
effect of industrial price rigidity on economic stability when agricultural 
prices are more flexible. The last question is not conclusively answered 
as yet, although a good many hypotheses, often based on hazy thinking 
or on misapplied partial equilibrium analysis, were tried and rejected. 
Adequate general analysis of income movements reveals the significance 
of such price inflexibility to be uncertain, or dependent on the relative 
weight of counterbalancing effects which are difficult to assess on a 
quantitative plane. 53 (One lesson to be learned from the controversy on 
this level is that the price theorist investigating prices will do well to 
suspect ad hoc norms in evaluating price results, and defer to genuine 
aggregative income analysis and to general equilibrium theory in this 
regard.) The discussion of the causes of price rigidity, beginning with 
Means' dichotomous distinction between administered and competitive 
prices, resulted in some needed elaborations of the theory of monopo- 
listic and oligopolistic price determination, until it is at present possible 
to tabulate a whole array of rationalizations for firms' pursuing relatively 
rigid price policies. However, applied analysis in specific cases has not 
been carried far enough fully to rationalize observed results in partic- 
ular industries. The statistical treatment of price inflexibility past and 
present has finally given us a fairly reliable measurement of the phe- 
nomenon, and an indication that inflexibility was not a sudden develop- 
ment of any particular decade. 51 The behavior of price over time should 
now assume a place commensurate to its moderate importance among the 
many significant results of industrial price policies. 

The preceding pages survey the general character of the findings of 
various sorts of empirical studies in the price policy field since the 
publication of Chamberlin's work. In spite of the aggregate resultant 
contribution, our empirical knowledge of price-making, its origins, and 
its results, is still very fragmentary. We have accumulated a general 
knowledge of market structures; a few suggestive but inconclusive meas- 
ures of demand and cost curves; a detailed analysis of pricing and com- 
petition in a few industries and a cursory survey of the same thing in 

53 See, for example, William Fellner, Monetary Policies and Full Employment (Berke- 
ley, 1946), pp. 137-140. 

54 A good general study of price rigidity is found in T.N.E.C. Monograph 1, Part I, 
Chap. 2 (S. Nelson, "Price Flexibility"). On p. 16 thereof, see a bibliography of litera- 
ture on the subject. 


a number of others; much undeveloped material on competitive behavior 
and tactics; a detailed analysis of price rigidity. Although the results of 
intensive individual studies are rich in suggestions, they are as yet too 
few, and their purposes are insufficiently unanimous, to permit much 
in the way of real comparative analysis, classification, or generalization. 

There has not as yet been any general and conclusive test of the 
predictions of abstract price theory. A good many empirical results are 
consistent with such theory and suggest that its assumptions are in 
general valid and its predictions correspondingly useful. But theoretical 
analysis often overestimates the sellers knowledge of governing data and 
his ability and disposition to calculate. As a result, actual price-making 
probably constitutes at best a very rough and often indirect approxi- 
mation, with a wide range of possible error, to the detailed predictions 
of geometrical analysis. 55 A further suggestion that may be elicited from 
empirical studies concerns the futility of attempting verification of over- 
simplified static theoretical models in a complex and dynamic real world. 

Empirical measurement and analysis is still too scattered to permit 
much in the way of inductive generalization concerning price-making, 
its results, and its origins. Various observers have advanced general 
hypotheses on the basis of fragmentary and imperfect evidence, and 
indeed some very tentative classifications and generalizations are sug 
gested by studies made so far. But any inductively derived explanatory 
systems are still definitely on the level of hypotheses tested only incon- 
clusively and by small samples. They are best viewed as tentative 
theoretical suggestions, and as such they will be discussed below. 

II. New Hypotheses, Issues, and Directions 
for Investigation 

Creative empirical work in any field will take its direction from major 
hypotheses advanced by those with especial insight and ability to grasp 
the broad possibilities of system in a type of behavior. But as it progresses, 
empirical study should reshape and elaborate its objectives by additional 
hypotheses which it develops. Price policy research is strongly indebted 
to the initial insights of a few people, and a fair measure of its vitality 
is its ability as it proceeds to turn up significant new issues for investi- 

Probably the most important single stimulus to price policy work has 
come from the broad and connected set of hypotheses, amounting to a 

c5 Cf. E. R. Hawkins, "Marketing and the Theory of Monopolistic Competition," 
Journal of Marketing, April 1940, IV, pp. 382-389. 


new theoretical system, advanced by Chamberlin. Additional emphasis 
of this fact is hardly required at this date. It is also evident that the many 
writings on the theories of monopolistic and imperfect competition since 
1933 have been a fruitful source of new hypotheses for empirical study. 56 
Equal attention at least must be given to those who in a sense "pioneered" 
in this related field of empirical investigation— who emphasized the 
close relation between what had previously been an area of institutional 
study and the new theory, who formulated problems for empirical in- 
vestigation, and who suggested the possibility of empirical generalizations 
which would do much more than test the simplified abstractions of a 
priori theory. In this regard major credit must be given to E. S. Mason 
of Harvard, who as much through the seminar room as through his 
writings 57 on price policy has influenced almost all workers in this field 
directly or indirectly. His ability to conceive the character and the 
exigencies of market analysis on the empirical level has been indeed 
striking. Especial importance may be attached to his suggestion that the 
explanation of price behavior be reduced to an objective level by going 
behind non-ascertainable demand curves to the observable characteristics 
of market structure, and by attempting to demonstrate the connection 
between market structure and price results. This suggestion has supplied 
a general basis for procedure in many industry studies. Other important 
influences on the development of a general attitude and approach to 
price policy research stemmed from A. R. Burns in his Decline of Com- 
petition and other writings, from Walton Hamilton, J. D. Black, and 
Donald Wallace. Their appreciation of the possibilities of research 
in this field, as much as their own empirical work, has been very im- 

A number of new hypotheses and issues have emerged from empirical 
work on pricing, and these are important not only as they affect further 
research but also for the suggestions they make for possible revision of 
price theory. Two of the most prominent suggestions concern the char- 
acter of the price-calculating processes of business firms and the question 
of whether the price-output results of various market situations are 
objectively determinate. 

Conventional a priori theory apparently takes a rather unequivocal 

66 Especial importance may also be attributed to such theoretical works as A. G. Hart, 
Anticipations, Uncertainty, and Dynamic Planning (Chicago, 1940); J. R. Hicks, Value 
and Capital (Oxford, 1939); and M. Abramovitz, Price Theory for a Changing Economy 
(New York, 1939)— all bearing on the problem of dynamic adjustments. 

67 See, for example, "Industrial Concentration and the Decline of Competition," 
Explorations in Economics (New York, 1936); and "Price and Production Policies of 
Large-Scale Enterprise," American Economic Review, March 1939, Proceedings, XXIX, 
pp. 61-74. 


stand on both issues. It suggests a definite process of calculating price, 
output, product, and selling cost, involving in general the finding of an 
optimum of maximizing positions where marginal cost is equal to mar- 
ginal revenue. It also implies, at least in the Chamberlinian version, that 
price-output results are in general, barring non-collusive oligopoly, ob- 
jectively determinate. That is, objective conditions of demand and cost 
govern behavior via the application of marginal calculation, and the 
subjective interpretations by sellers of these conditions are not so un- 
certain or random as to require abandonment of the assumption of 
uniquely and objectively given data. 

The first point in itself has become the subject of much controversy. 
Many empirical studies suggest that business managers do take account 
of the demands for their own outputs, or, on the supposition of con- 
current price policies by rivals, of their "industry" demands. It has also 
appeared that they are aware of their costs and of the manner in which 
these change with changes in output and product, and that they consider 
the relation of selling cost to demand. Such studies often also suggest, 
however, that the number of variables and functional relationships gov- 
erning enterprise profit is great, especially when many of the variables 
are dated, and furthermore that the magnitudes of many relevant data 
are uncertain. And they suggest that all of this does result, at least in 
certain industries, in the employment of "crude" formulae or approxi-' 
mation methods of determining price, output, product, and selling costs, 
and that these may not result in precise equation of marginal cost to 
marginal revenue (or in whatever is equivalent to this as a means of 
maximizing profits in multi-variate models). 

Controversy has flared when some have suggested that a marginal 
calculation is not followed at all, at least in oligopoly, and that pricing 
on the basis of full average cost plus a margin may constitute a more 
general rule. 58 In response to these insidious suggestions, "marginalism" 
has been avidly defended. 59 

In some perhaps uncommon cases, of course, for example in oligopoly 
with no express or tacit collusion, "full-cost" or kindred formula pricing 
may be the occasion of no legitimate controversy, since in the absence of 
a uniquely determined marginal revenue curve for the seller, employ- 
ment of an arbitrary formula may involve no contradiction of theoretical 
predictions. In all other cases, use of genuinely arbitrary formulae might 
be surprising, although use of rough approximation should not be, just 

68 See, for example, R. L. Hall and C. J. Hitch, "Price Theory and Business Behavior," 
Oxford Economic Papers, No. 2, May 1939, pp. 12-45. 

59 See F. Machlup, "Marginal Analysis and Empirical Research," American Economic 
Review, September 1946, XXXVI, pp. 519-554. 


as management should not be expected to conceive and solve a multi- 
dimensional problem in terms of an artificially simplified two-dimensional 
model occurring in economics textbooks. Empirical research, in any 
event, has not as yet established a major probability that any arbitrary 
formula is the key to price making in any market category. It has left 
open the possibility that flexible application of such formulae may give 
results roughly equivalent to marginal calculations, and it has suggested 
many apparent alternatives to straightforward "full cost" pricing. The 
facts here, and their interpretation, are still largely to be discovered. 

What this research does suggest positively is the following (and this 
quite aside from any potential oligopolistic uncertainty): (i) The ap- 
plicable theoretical models for profit maximization are much more com- 
plex, because of the number of variables, relationships, and time periods 
involved, than those ordinarily seen in textbooks. No simple verification 
of two-dimensional period-by-period balance of cost and marginal reve- 
nue should ordinarily be expected by the most doctrinaire proponent of 
"marginalism"; unless these two functions are implicitly redefined to 
reflect all forces bearing on the problem (and thus made intangible if 
not meaningless), profit maximization does not require that they be 
equated in every period. (2) If theory is taken in an applicable multi- 
dimensional version, the enterpriser's ability to carry out the implied 
profit-maximizing calculations can easily be overestimated. (3) The 
enterprisers knowledge of governing data and functional relation- 
ships also tends to be overestimated in a priori theory which as- 
sumes given data. The range of uncertainty concerning many variables 
and relationships may be very great— in fact great enough to discourage 
precise handling. (4) In view of all this, we may question not only the 
ability but also the disposition of sellers to make highly complicated 
calculations with highly uncertain data, and may understand it as per- 
fectly rational if they employ various simplified formulae and approxi- 
mation methods in arriving at a desirable price and output. 

So far as it intends to make objective predictions of behavior in ob- 
jective situations, price theory must take some of these considerations 
into account. The task of empirical research would seem to be to find out 
in many more cases how prices are made, and to examine the price-mak- 
ing process to find if the results are highly systematic, or in fact rather 
arbitrary and capricious. It is not evident that the question of whether 
"profits are maximized" is really amenable to treatment. 

All of this bears on the issue as to whether competitive behavior and 
price results are objectively determinate. There has been a thought- 
provoking exchange of views on this matter, principally between E. S. 


Mason and E. G. Nourse. The essence of Mason's position is that there 
may be a determinate association between objectively ascertainable char- 
acteristics of market structure and price results. Accordingly he advanced 
the hypothesis that the objective market situation is primarily determin- 
ing, in the sense that different individuals placed in the same situation 
would make price policy decisions in approximately the same way. 60 
(It will be noted that this is the same sort of objective determinism 
which Chamberlin would suggest, except that it emphasizes as deter- 
minants ascertainable structural characteristics, such as the numbers 
of sellers and buyers, instead of the largely non-ascertainable individual- 
seller demand curves.) Nourse's position, as developed in two books 61 
and other writings, is that with concentrated markets, large corporate 
firms, and the complex structure of modern entrepreneurship, competitive 
behavior is far from determinate. He holds that there is often a broad 
scope for creative decision-making by the business executive. In effect, 
different persons, each desiring to "maximize profit," might act in very 
different ways in the same objective situation, because they might 
interpret its possibilities differently, might see in different lights the 
feasibility of altering the market situation, or might be optimistic in 
different degrees. This is held to be especially true because of the wide 
range of discretion open to executives with regard to making innovations 
of technique or product, and because of the multiplication of uncertainty 
in situations which are dynamic in these dimensions. 

The resulting issue has scarcely been resolved. Any hypothesis of 
objective determinism is in danger of overlooking the fact that human 
behavior and not simply equation-solving is involved, and of viewing 
business executives as possessed mainly of a sort of standard profit-maxi- 
mizing reflex which automatically finds the correct implications of a 
given market setting. This view may be misleading if we recognize 
that the controlling variables are many, that some of their values are 
highly uncertain, that the ability and disposition of executives to solve 
highly complicated maximization problems employing guesswork data 
may be limited, and that in genuinely non-collusive oligopolistic situ- 
ations the controlling data for the firm may not even be hypothetically 
determinate. In view of these considerations, significant differences in 
behavior might be expected to result from substituting different persons 
in given objective situations, and the fact that these situations are defined 
in empirically tangible and measurable terms in no wise alters this 

&) E. S. Mason, "Price and Production Policies of Large-Scale Enterprise," loc. cit. 
61 E. G. Nourse and H. B. Drury, op. cit.; also E. G. Nourse, Price Making in a 
Democracy (Washington, 1944). 


probability. The question remains, however, whether behavior may not 
very frequently be determinate within a relatively narrow range, so that 
determinism is saved after all, and Mr. Nourse's creative executive is 
ordinarily confined to operations within narrow limits. The answer may 
lie in an extended comparative analysis of the uniformities of executive 
behavior in parallel situations. It will also be pertinent to observe, in the 
developing national policy situation, the extent to which business "profit 
maximization" becomes, through the recognition of all sorts of secondary 
and tertiary consequences of pricing action, an entirely plastic and in- 
definite concept. The writer suggests tentatively (i) that any given 
"static" market structure ordinarily confines executive discretion to a 
relatively narrow range (as long as, in oligopoly, he observes the usual 
tacit or explicit collusion on price), so that meaningful associations be- 
tween quasi-static market structure and competitive behavior can often 
be established for moderate time intervals, but (2) that the executive 
has enough discretion at any one time that he may follow significantly 
different policies designed to change market structure through time, and 
(3) that the dynamic course of market structure (and hence behavior) 
over substantial time intervals may not be at all determinate. 

Empirical research has also made a contribution to the general analysis 
of price in the area of geographical price policy and spatial price differ- 
entials. Detailed analyses of basing-point systems, of the sort developed 
by de Chazeau 02 in his work on steel, together with various controversial 
discussions of basing-point pricing, 63 have developed, in terms congruent 
with theories of oligopoly and monopolistic competition, an economic 
analysis of the origins of and alternatives to systems of geographical 
price discrimination. Strictly a priori analysis has of course shared the 
burden of this development. As a total result, we have a more adequate 
interpretation of geographical pricing phenomena. 

An issue still worth attention was introduced, or revived in a new 
context, by the writing of those persons connected with the anti-trust 
law enforcement after 1937. 64 This concerns the character and impor- 
tance of collusive activities in price and output determination in our 

62 C. R. Daugherty, M. G. de Chazeau, and S. S. Stratton, op. cit., Ch. 12-14. 

63 Such as F. A. Fetter, "The New Plea for Basing Point Monopoly," Journal of 
Political Economy, October 1937, XLV, pp. 577-605; M. G. de Chazeau, "Reply to 
Professor Fetter," ibid., August 1938, XLVI, pp. 537-566; A. Smithies, "Aspects of the 
Basing Point System," American Economic Review, December 1942, XXXII, pp. 705- 
726; V. A. Mund, "Monopolistic Competition Theory and Public Price Policy," ibid., 
pp. 727-743; J. M. Clark, "Imperfect Competition Theory and Basing Point Pricing," 
ibid., June 1943, XXXIII, pp. 283-300; and numerous others. 

64 See, for example, Thurman Arnold, Bottlenecks of Business; Corwin Edwards, "Can 
the Anti-Trust Laws Preserve Competition?" American Economic Review, March 1940, 
Proceedings, XXX, pp. 164-179. 


predominantly oligopolistic economy. In the first flush of enchantment 
with the new price theory, and of reaction against the old "trust problem" 
study, there may have been some tendency to discount the importance 
of collusion. It may have been supposed that the theory of oligopoly 
pricing showed collusion to be unessential— "mutually recognized inter- 
dependence" taking its place or giving the same result— or that emphasis 
on the firm's demand curve eliminated the necessity of direct reference 
to crass institutional matters. Such an emphasis involves arid formalism 
and perhaps a flight from reality into the calculus. The suggestions of 
Arnold and Edwards were thus useful in reminding us: (1) that col- 
lusion in some sense often if not commonly plays a strategic role in the 
process of price formation; (2) that it is a very complex phenomenon, 
and can assume many significantly different forms; (3) that price be- 
havior may vary with the sort of collusion adopted, and with the state 
of law and law enforcement affecting collusion. Such suggestions tend 
to undermine somewhat the notion that certain price results are inevi- 
tably associated with given market structures, unless indeed these market 
structures are defined to include in a given state all relevant aspects of 
the legal framework controlling or influencing competitive and collusive 
behavior. They imply that the theory of oligopoly price should put much 
more emphasis on collusive models, 65 and that empirical research should 
attempt to deal in some detail with the nature and effects of various types 
of collusive arrangements. 

If empirical research has cast some light on the theoretical issues like 
those just described, it has made little definite progress as yet on a 
larger problem— that of establishing an objective classification of markets, 
each sub-category of which would contain industries with a uniform 
and distinctive type of competitive behavior. Such a classification is of 
course a goal of empirical generalization in the price policy field, if indeed 
such generalization is possible. Classifications so far developed appear 
to be substantially non-objective in character, or to be too general to 
allow separation of distinctive types of behavior, or, if more detailed, to 
represent imperfectly substantiated guesses requiring more extended 
empirical backing. "Degree of monopoly" classifications in the Robin- 
sonian tradition, like those advanced by Lerner 66 and Rothschild, 67 
rest upon practically non-ascertainable sellers' demand curves and can- 

66 Cf. D. Patinkin, "Multiple-Plant Firms, Cartels, and Imperfect Competition," 
Quarterly Journal of Economics, February 1947, LXI, pp. 173—205. 

06 A. P. Lerner, "The Concept of Monopoly Power," Review of Economic Studies, 
June 1934, Ii PP- 157-175- 

67 K. W. Rothschild, "The Degree of Monopoly," Economica, February 1942, XXIV, 
pp. 24-39. 


not be considered as objective or empirically applicable. 08 The Chamber- 
linian market classification, based simply upon a twofold classification 
of product difference and a threefold classification of number of sellers 
(one, a few, and many), 09 is objective but evidently too general. In 
particular, it leaves the oligopoly category, which in fact comprehends 
the great majority of actual cases, as a substantially undifferentiated 
and amorphous mass. The need is particularly evident for finding first 
whether other market characteristics, besides number of sellers and prod- 
uct differentiation, offer a basis for distinguishing (at least qualitatively) 
particular types of oligopolistic behavior, and second what the effect 
may be, within the general bounds of oligopoly, of differing degrees 
and patterns of concentration. 

Empirical research, industry by industy or otherwise, simply has not 
proceeded far enough to give entirely reliable general suggestions on 
either of these points. There is really not enough accumulated infor- 
mation to permit the conclusive establishment of an explanatory classi- 
fication which will account even on a qualitative level for observed 
differences in oligopolistic behavior, or, conversely, to support the con- 
clusive rejection of the possibility of such a classification. And there are 
ample reasons for wondering if any such classification with demonstra- 
ble explanatory value can be developed in a world where the effects of 
dynamic change of and random uncertainty concerning the governing 
data may obliterate or obscure the virtual influence of basic environ- 
mental conditions. Nevertheless, various "scattered returns" have been 
drawn upon to formulate some tentative market classifications which 
go beyond those found in a priori theory. Even though these classifi- 
cations represent casually tested hypotheses rather than established 
findings, a mention of them may be deserved, in part because they have 
provided a basis for the experimental elaboration of the assumptions of 
a priori price theory. 

J. M. Clark, in his notable article on workable competition, 70 has 
singled out as potentially important, in addition to number of sellers 
and product differentiation, such things as the size distribution of pro- 
ducers, the methods of price-making and selling, the perfection or im- 
perfection of market organization, and spatial relationships. Recognition 

G8 The Triffin classification (Robert Triffin, Monopolistic Competition and General 
Equilibrium Theory [Cambridge, Mass., 1940]), resting on the cross-elasticities of sellers' 
demands, has a similar limitation. 

69 See F. Machlup, "Monopoly and Competition," American Economic Review, Sep^ 
tember 1937, XXVII, pp. 445-451. 

70 "Toward a Concept of Workable Competition," American Economic Review, June 
1940, XXX, pp. 241-256. 


of these characteristics in classifying markets might give rise to a tentative 
classification (abbreviating Clark's own) of the following general form: 

I. Pure competition— standard product, known (quoted or supply-governed) price, 
many sellers at any local market, free entry. 

A. Perfect competition (perfect factor mobility). 

B. Imperfect competition (imperfect factor mobility). 

i . Excess capacity (price less than average cost on average over time). 
2. No excess capacity. 
II. Modified, intermediate, or hybrid competition. 

A. Standard products, few sellers, free entry but exit with loss. 

i . Quoted price, without significant spatial separation of producers. 

a. Open price. 

b. Imperfectly known price, chaotic discrimination. 

c. Open price with limited or occasional departures. 

2. Supply-governed price (open market). 

3. Quoted prices, with significant spatial separation of producers. 

B. Unstandardized products, either many or few sellers. 

1. Quoted prices. 

2. Supply-governed prices. 

With the omission of numerous asides on the assumed shape of the 
individual seller's demand function in various sub-categories, this is 
the essence of Clark's classification so far as it runs in terms of ascertain- 
able market characteristics. The state of current knowledge is reflected 
in the fact that although this classification may contain very pertinent 
suggestions, we cannot say with certainty whether or not Clark has hit 
upon the most essential characteristics distinguishing among markets 
within the oligopoly category or among markets generally. The writer 
would object strongly to lumping all differentiated-product industries in 
one category, regardless of number of sellers, and to neglecting the 
number of buyers, the durability of output, the difference between rela- 
tively easy and very difficult entry in oligopoly, and the time-trend of 
industry demand. And the extended distinctions among quoted and 
"supply-governed" prices in oligopoly seem to receive more emphasis 
than these actually tenuous distinctions may deserve. With Clark we 
would have automobiles, rubber tires, cigarettes, light bulbs, ladies' 
dresses, radio sets, optical goods, agricultural equipment, and electrical 
machinery nested incompatibly together in category II-B-i, although 
there are significant ascertainable and clearly explicable differences in 
competitive behavior and price results among these industries. But much 
more investigation would be required to evaluate Clark's classification 

The penchant of this writer, after the suggestions of Mason and 


Wallace in their work at Harvard, would be to emphasize the following 
market characteristics as strategic in explaining observed differences in 
competitive and price behavior: 71 the number and size distribution of 
sellers, the number and size distribution of buyers, whether a producer 
or consumer good (linked to product differentiation), the conditions of 
entry, the durability of the product, the time-trend of industry demand. 
From this could develop an abbreviated classification in the following 
form : 

I. Many sellers, free entry. 

A. Consumers' goods, differentiated products, many buyers. 
i. Durable and style-varied goods. 

2. Non-durable goods. 

B. Producers' goods, unimportant product differentiation (not distinguished on 
basis of durability). 

i . Many buyers. 
2. Few buyers. 

(Further distinction possible on the basis of time-trend of demand.) 
II. Few sellers in general. 

A. Consumers' goods, differentiated products, many buyers. 

i. High concentration of output in hands of few sellers, very difficult entry. 

a. Durable goods, strong style elements. 

b. Non-durable goods. 

2. Moderate concentration, relatively easy entrv (not distinguished on basis 
of durability). 

B. Producers' goods (not distinguished on basis of durability or product dif- 

i. High concentration, difficult entry. 

a. Many buyers. 

b. Few buyers. 

2. Moderate concentration, relatively easy entry. 

a. Many buyers. 

b. Few buyers. 

(Further distinction on the basis of time-trend of industry demand desir- 

This classification, a revision of a basic model introduced by E. S. 
Mason about a decade ago, seems to the writer to establish sub-categories 
within each of which the available evidence (catalogued in the pre- 
ceding section) points to significant uniformities of behavior and among 
which there are systematic and significant differences, with respect to 
such matters as profit rates, ratio of selling to production cost, price be- 
havior over time, product behavior over time, and so forth. Discussion 
of these uniformities and differences would require more space than 

71 See J. S. Bain, "Market Classifications in Modern Price Theory," Quarterly Journal 
of Economics, August 1942, LVI, pp. 560-574. 


here available, but in any event it must be emphasized that these are 
only very tentatively indicated and by no means conclusively established, 
The problem of explanatory classification thus remains open, and an 
explicit orientation of research toward its solution would be highly 
desirable. Experimentation with abstract price theory employing corre- 
spondingly elaborated assumptions, together with some realistic assump- 
tion concerning collusion, would also be desirable. 72 

These are some of the theoretical issues raised by recent price policy 
research. Although the stimulus to abstract analysis may be considerable, 
progress in the development of empirical generalities to resolve them is 
quite slow. The lagging pace of empirical discovery and generalization 
may be unavoidable, but it does raise primary questions concerning the 
method of research. What should research workers be doing and how 
should they be trying to do it? On these points there is much indecision, 
lack of direction, and difference of opinion. It may therefore be appropri- 
ate to consider the problem of method in this area of research. 

III. Aim and Method in Price Policy Research 

The variety of types of empirical price study reviewed earlier in this 
discussion suggests that there is as yet no settled agreement on specific 
objectives of research or on the method of analysis. At this stage, of 
course, there is no strong case for standardization of method; the field 
is new and much of the work necessarily involves exploration in methods. 
Some conscious recognition of alternatives in aim and method, and some 
weighing of their relative potentialities, may nevertheless be desirable. 

There are several distinct but potentially allied aims of empirical re- 
search into pricing: (i) description and fact-finding, (2) interpretation 
and generalization, and (3) normative evaluation and critique. Several 
things may be emphasized about description and fact-finding. A con- 
siderable contribution to knowledge can result from comprehensive and 
detailed findings of the facts concerning market structures, competitive 
behavior, the processes of price formation, and the price, output, and 
allied results which emerge. The finding of facts, on any or all of these 
levels, is an indispensable part of effective empirical research, and con- 

72 As of the end of the period here surveyed, a well-developed and novel theoretical 
system has been put forth, which although it has not as yet obviously influenced price 
policy research, may eventually have some impact. This is the Neumann and Morgen- 
stern work on the Theory of Games and Economic Behavior (Princeton, 1944), which 
suggests a new approach to the formation of business decisions under conditions of 
recognized interdependence. What influence it may have on the interpretation of price 
policies will be interesting to observe. 


siderable debt is due to studies which in effect go no farther than this. 
But there is a question how much such facts per se contribute to our 
understanding of economic activity and its consequences. Even an indefi- 
nitely large compilation of facts, selected in accordance with some general 
criteria of relevance, may not lead at all automatically to effective gen- 
eralization about or evaluation of the behavior in question. Fact-finding 
and measurement are in effect not desirably isolated from interpretation, 
explanation, and evaluation. Serious scientific generalization is the prime 
responsibility of any worker in this field; the production of systematic 
statistics and similar data contributes to, but is only a part of, this 
process. More important, adequate criteria of relevance cannot be main- 
tained and adequate direction cannot be given processes of definition 
and measurement unless the person charged w 7 ith organizing the facts 
is also well informed concerning the detailed problems of interpretation 
and evaluation. And we certainly should not confuse an unsystematized 
catalogue of facts with any real understanding of the pricing process. 

Interpretation and explanation, proceeding from observed association 
in the specific case into empirical generalization comprehending many 
or all cases, is the appropriate central aim in price policy research. If our 
knowledge is to be operationally useful, we need to know what is system- 
atically associated with what— for example, what the demonstrable 
association is among market structure, competitive behavior, and price 
results. Factual investigation should be oriented to the exigencies of 
generalization, and should be tested for its usefulness in this regard as 
rapidly as possible. 

The attempt to develop empirical generalizations raises the question 
of method. Without attempting to exhaust the subject, we may note that 
there are two sets of choices with respect to method concerning which 
workers in the field seem to differ. 

The first choice involves the degree of dependence upon the broad 
hypotheses and explanatory mechanisms of the existing a priori price 
theory. Three possible approaches are available here: (i) the attempt to 
verify (or disprove) the predictions of price theory in an explicit and 
quantitative sense, by measuring the presumably determining variables 
and functional relationships, and by checking their implications against 
measured results; (2) the attempt to establish, and possibly to elaborate 
and revise, the broad qualitative predictions of such theory, but without 
explicit dependence upon statistical ascertainment of determining func- 
tions; and (3) a strictly experimental attempt to find any sort of system- 
atic associations, with little or no dependence upon conventional theory. 
There are examples of each approach. Joel Dean's statistical cost curve 


work seems related to the first; much of the general "industry study" ap- 
proach espouses the second; Walton Hamilton is perhaps the most ex- 
plicit exponent of the third. 

If it is to do more than indicate the probable shapes of some of the 
determining functions employed in abstract analysis, the first approach 
labors under severe difficulties. Simplified textbook price theory does not 
ordinarily provide an adequate analytical framework for verification, and 
a sufficiently elaborated, multi-variate theory would require discourag- 
ingly complex statistical analysis. In numerous cases, the strategic indi- 
vidual-seller demand functions may not be susceptible of objective 
ascertainment, or if they are to be found, we must turn the subjective 
estimates of sellers into statistical data. The feasibility of this procedure 
may certainly be questioned, especially if sellers in complex and uncer- 
tain situations adopt short-cut formulae and do not deal explicitly with 
the theoretically controlling functions. Finally, the uncertainties of corre- 
lation analysis as employed in all these pursuits reduce the significance of 
the findings that emerge. For these reasons, effective empirical general- 
ization on a quantitative level concerning the association of price results 
with determining variables seems unlikely to be attained. Perhaps it is 
better to regard price theory as an embarkation point and general guide, 
rather than as a fixed body of principles for detailed verification. A gen- 
eral commentary on and elaboration of conventional theory may have to 
be accepted as a sufficient accomplishment in this direction. 

A somewhat less ambitious approach, following the general suggestion 
of price theory, attempts to establish systematic associations among easily 
ascertainable characteristics of market structure, patterns of competitive 
behavior and price policy, and observable price results. 73 This is indeed 
the general method common to a large number of empirical studies in the 
price policy field. It does not necessarily deny the implicit importance of 
governing functional relationships, but expediently chooses to adhere to 
what can be seen, measured, and exhibited, thus placing any generaliza- 
tions which may emerge on a substantially objective level. 

Studies following this pattern suggest that it has the advantage of flexi- 
bility while still recognizing some logical order. In particular, it need not 
be tied to the static reference implied in verification of conventional price 
analysis, and can deal directly with evolution as well as static equilibrium 
tendencies. Moreover, while able to admit a? many ascertainable deter- 
minants as necessary, it is not committed to ihe defense of a fancifully 
complex maze of hypothetical governing relationships. But it has also 
very clear limitations. Such generalizations as the method develops have 

73 See E. S. Mason, "Price and Production Policies of Large-Scale Enterprise." loc. cit. 


been and promise to be largely on a qualitative level. (The association 
between blockaded entry and large profits may be established, but hardly 
that between a certain "degree of blockading" of a certain percentage 
return on invested capital.) And simple associations are unlikely to be 
established— as between concentration and price rigidity— in a world 
where many dimensions of market structure bear on any single result. 
The demonstrable associations in each individual case may in tact be so 
complex that the scope of generalization is severely limited. Without the 
greatest perception and insight, research of this sort may lead only to 
analytical economic history for a catalogue of individual cases. This 
would be useful, but it would not reach the goal of generalization. It is 
therefore a very serious question whether price policy research can move 
with finality above the first level and gain a real foothold on the second. 

A third approach recognizes no specific debt to the concepts and gen- 
eral hypotheses of price theory, and proceeds empirically to catalogue 
cases, perhaps in the hope of learning if regularities of any sort can be 
extracted. This attitude might presumably be justified by the supposition 
that the hypotheses of price theory are relatively useless or unduly restric- 
tive, and that an honest and careful empirical study beginning from 
scratch will be at least as productive as any other. Unless the possibility 
and usefulness of generalization are also denied, the fruitfulness of such 
an approach will be tested by its demonstrated ability to arrive at any 
generalizations or explanations whatever. And, on the basis of current 
evidence, the validity of the more general hypotheses of conventional 
price theory does not seem so discredited as to recommend that they be 
abandoned indiscriminately. 

So much for the general approach to interpretation and explanation of 
business behavior. A second choice in method concerns the appropriate 
scope of the individual investigation. Should we select the firm or indus- 
try as a focus for intensive investigation, or can meaningful results be 
extracted from a broad cross-section study of a large number of firms or 

If the student intends to verify or disprove the constructions of conven- 
tional price theory, the appropriate focus is first the firm— to be studied 
intensively and in detail— and second the industry. Really detailed 
analysis at this level should be required. Supposing that the defense or 
refutation of conventional theory or the development of a substitute is 
contemplated, the Oxford game of Twenty Questions, as played by Hall 
and Hitch and by Saxton 74 with a random sample of thirty or forty busi- 
nessmen, is not a promising pursuit. Broad and superficial questionnaire 

"'* C. Clive Saxton, Economics of Price Determination (Oxford, 1942). 


investigations into pricing methods may raise many challenging ques- 
tions, but they provide conclusive or reliable answers to practically none 
of them. Attempts to generalize from such casual findings are hardly to 
be taken seriously. 

Is the conclusion different if the aim is to develop empirical generaliza- 
tions of a qualitative sort, following the broad hypotheses of theory but 
not attempting to verify its reasoning in detail? Again the broad and rela- 
tively superficial studies, depending upon a limited range of information 
from each of a number of cases, have been disappointing. Suppose that 
we take 50 manufacturers in a variety of industries and market situations 
and ask them all, as Saxton 75 does, questions such as: "Do you find resist- 
ance to price changes from wholesaler, retailer, and/or consumer? ... 
If demand increases sharply . . . would you increase selling prices 
(under various alternative cost conditions)?" And suppose we find that 
on the second question 29 answer "no," 6 "yes" and 15 "maybe" (de- 
pending on cost conditions). It is not apparent that by any number of 
such questions so administered we gain very much systematic knowledge 
of price policies, and it is certain that we learn little if anything at all 
about their origins or their consequences. Intensive investigation of fewer 
cases, not relying primarily on diffuse questionnaire data but probing at 
length into the basic environment of decision-making and into the objec- 
tive consequences of these decisions, would seem to be a prime requisite 
for the development of meaningful or conclusive findings. In such studies, 
an appropriate employment of questionnaire and interview methods 
might assist considerably in gaining a knowledge of those managerial mo- 
tivations and price-calculating processes which logically bridge the gap 
between observed environment and observed price and output results. 
The questionnaire technique fer se has been disappointing in this field to 
date mainly because of the way in which it has been employed. 

Related limitations have been encountered in attempts at generaliza- 
tion through studies involving broad correlations of published statistical 
data. Thorp's analysis of the (slight) relation between price rigidity and 
industrial concentration 76 was salutary in refuting some oversimplified 
hypotheses concerning this phenomenon. But its negative results sug- 
gested strongly that in a field of complex multi-variate relationships, the 
associations which can be tested from unrefined published data are few 
indeed. It is a bothersome fact that the bulk of relevant data for any firm 
or industry must be developed and refined by rather intensive investiga- 
tion of that unit. The generally published data will not support much in 

75 Il7icJ., pp. 182, 183. 

76 W. L. Thorp and W. F. Crowder, of. cit. 


the way of meaningful correlation procedures, and there is no fast and 
easy shortcut to the information needed. 

The most promising results we have to date— in the way of reasonably 
complete and defensible demonstration of possible causative associations, 
and of convincing appraisal of policies and their results— are found in the 
individual firm and industry studies, and principally in the more detailed 
and intensive ones. The method of the extensive case study, frequently 
repeated, seems the best yet discovered for this field. But each case study 
takes a long time, and many of them are required to permit effective gen- 
eralizations. The prospect of getting general results in the near future is 
not promising. 

IV. The Development of Norms of Price Behavior 

Ideally, an integral part of research into price and production policies 
should be the evaluation, in terms of some sort of norms, of the price and 
related results to which these policies give rise. Description and interpre- 
tation of competition and its results are certainly desirable, but they find 
a justification beyond the simple pursuit of knowledge in the ultimate 
possibility of evaluating the efficiency with which the enterprise system 
utilizes resources. The more satisfying studies of price policy have in fact 
carried through from description and explanation of price policies and 
results to a critical evaluation of these results. 77 

Any account of developments in the price policy field should thus refer 
to progress in the development and application of norms of behavior. The 
process of normative evaluation of industrial behavior should generally 
involve three steps: identification of certain strategic dimensions of ag- 
gregate economic welfare which may be influenced by the price-output 
results of individual industries; establishment of causal interrelationships 
between specific price-output results and specific dimensions of the aggre- 
gate welfare; and definition of normative values for price-output results, 
either individually or as an interrelated complex. Thus we might identify 
the level of employment as strategic and a certain level as most desirable, 
establish a presumed relation between the ratio of profits to other income 
shares and the level of employment, and then attempt to define an ideal 
magnitude for profits in a given situation. Or we might define a hypo- 
thetically ideal allocation of resources, establish a connection between 
desirability of allocation and the price-marginal cost ratios or the selling 
cost-production cost ratios of industries, and attempt to define ideal ratios. 

77 It may be noted that normative standards can be meaningfully applied to any 
results which have been measured, even though their causes are imperfectly explained. 


Similar sequences of reasoning might be developed to deal with cyclical 
stability, efficiency, progressiveness, and so forth. 

The principal progress in the normative field during the period in 
question is found in the development of general economic analysis to the 
point where the existence of strategic relationships can at least be readily 
recognized and the idea of scientific material welfare norms of individual- 
industry price behavior introduced. Two developments deserve especial 
mention: first, the theory of general relative-price equilibrium and the 
underlying analysis of consumer choice and satisfaction have been ex- 
tended to provide tentative norms of allocation in an economy of quasi- 
monopolistic firms (especially in work of Lerner and Kahn); and sec- 
ondly, the Keynesian analysis of the determinants of employment has 
opened avenues for linking profit and price behavior with the level and 
stability of employment. With this fundamental beginning, a few writers 
have addressed themselves forthwith to the task of developing specific 
norms for price-cost relationships and similar individual-industry results 
in terms of their impact on employment, allocation, efficiency, progres- 
siveness, stability, and the like. 78 The net resulting accomplishment to 
date is somewhat difficult to characterize. At a minimum, it has been 
demonstrated clearly that appropriate criteria for evaluating behavior can 
be drawn from an economic analysis which provides an explicit mecha- 
nism for tracing the impact of individual business actions on dimensions 
of the total welfare which commonly accepted value-judgments hold to 
be important. And it has thus been possible to envisage transcending a 
primary dependence either on the norms of law or on simple ad hoc judg- 
ments devised for the occasion. Beyond this, some very tentative norms 
have been advanced, at least for certain types of price result. But these 
norms so far have not been conclusive or even especially helpful in prac- 
tice—at any rate not much more so than a good investigator's ad hoc 

The primary difficulties are found in the lagging development of 
"pure" economic analysis to supply the mechanism for developing norms. 
The Keynesian analysis, in its current state of development, places very 
uncertain limits, for example, on the desirable magnitude for profits. In- 
terpretations of business cycles are sufficiently various and complex as to 
leave us without much guide on the desirable degree of price flexibility. 
Norms of allocation and income distribution derived basically from an 
analysis of purely competitive economies can be applied to an economy of 

78 D. H. Wallace, "Industrial Markets and Public Policy," in Public Policy, Harvard 
Graduate School of Public Administration (Cambridge, Mass., 1940), pp. 59-129; and 
J. M. Clark, "Toward a Concept or Workable Competition," loc. cit. 


quasi-monopolistic markets only with adaptations which arc as yet imper- 
fectly developed. The evaluation of an economy in process of dynamic 
change in terms of norms drawn from the theory of stationary equilib- 
rium is evidently unsatisfactory, as Professor Schumpeter has pointed 
out 79 (suggesting the rejection of much orthodoxy concerning static equi- 
librium norms for price behavior), but dynamic process analysis is as yet 
not sufficiently elaborated to give us an adequate substitute. And we have 
no adequate theoretical basis for developing norms of selling costs and 
non-price competition. This catalogue of obstacles may serve to empha- 
size one essential fact— the development of satisfactory norms of industry 
behavior is a task not of empirical investigation but of pure theory. No 
matter how adroit the student of price policy may be in applying received 
doctrine to the normative problem, he will be, until there are substantial 
advances in pure theory, in the position of shoveling down a mountain 
with a teaspoon. 

This is to say that we neither have nor have in sight an adequate and 
dependable set of norms of satisfactory price-output results for individual 
industries. We are thus even farther from norms of satisfactory market 
structures. It is of course possible, while leaning heavily on undesirable 
ceteris paribus clauses, to make some general qualitative judgments con- 
cerning the relative desirability of certain results : obvious redundancy of 
plant is wasteful; huge competitive selling costs are questionable; very 
large profits gained without compensating advantage distort income dis- 
tribution; and persistent suppression of product innovation is suspect. But 
in evaluating the general-equilibrium consequences of the particular- 
equilibrium behavior of an industry or group of industries, our theory 
permits us to make judgments mainly on the broadest and most obvious 
level, or, in effect, on a level which could be approximated fairly well 
with lay common sense. 

In the industry studies reviewed earlier, the authors have been forced 
to rely for norms either on the rather inadequate theoretical models on 
hand or on makeshift substitutes. The element of improvisation is espe- 
cially strong when they turn to the evaluation of selling costs, price in- 
flexibility, or geographical price patterns; and blind faith seems to prevail 
when they turn to the problem of the allocation of resources. The alle- 
giance to norms derived from an analysis of static equilibria is particularly 
apparent in a number of the studies mentioned. This is not said in dis- 
paragement of the work, as investigators have had to work with the tools 
at hand, and much of the improvisation has been of a high order. But 

79 J. A Schumpeter, Capitalism, Socialism, and Democracy \ov York, 1942), 
Part II. 


there is a clear need for a general development of a basic theory and of 
norms both appropriate to a quasi-monopolistic economy in dynamic 
process, if evaluations of industrial price behavior are to have a certain 
and dependable reference. 

V. The Influence of Price Policy Research on Attitudes 
Toward Public Policy 

Chamberlinian and allied theory, and much of the price policy investi- 
gation related to it, have naturally influenced thinking on public policy 
toward competition and pricing. It would probably be too much to sug- 
gest that a new orthodox view of the proper ends and means of govern- 
ment regulation has emerged, since there is still much disagreement 
about both ends and means. Moreover, the bulk of "policy thought" over 
a considerable period has centered on fiscal and monetary matters, and 
has tended to neglect or regard as relatively impractical attempts to influ- 
ence pricing and competition. Certain positive influences on policy think- 
ing, however, are apparent. 

The theory of monopolistic competition was salutary in focusing atten- 
tion on the various specific price-output results potentially associated with 
various sorts of market structure. When extended by an adequate norma- 
tive evaluation of such results, necessarily involving reference to general 
and aggregative analysis, this theory may ultimately assist us in suggest- 
ing explicit criteria as to (1) where government intervention is desirable, 
and (2) what the ends of this intervention should be. A much more ex- 
plicit orientation for policy, or at any rate one more clearly linked to 
broad goals of welfare maximization, was thus introduced. 

A second possible suggestion of the theory of monopolistic competition 
was that observed price-output behavior is relatively determinate by virtue 
of the independent action of sellers within the given market structures, 
and hence inevitable unless directly regulated. This is hardly a legitimate 
inference from the theory, but it was frequently drawn and has led to 
suggestions concerning the a priori futility of anti-trust prosecutions of 
collusion as a means of influencing behavior. It also led to proposals for 
direct price regulation as an alternative. 80 This was largely a transitional 
attitude, however. The arguments of Arnold and Edwards concerning 
the importance of collusion, and of Nourse concerning the potential vari- 
ability of price policy, have shaken earlier convictions that price results 

60 See W. H. Nicholls, "Social Biases and Recent Theories of Competition," Quarterly 
Journal of Economics, November 1943, LVIII, pp. 1-26, for a related but much more 
extended discussion of this issue. 


are strictly determinate regardless of collusion (or even with collusion) 
and that collusion and procedures directed against it are unimportant. At 
the present time the contributions of the new theory are mainly in facili- 
tating the introduction of criteria of economic analysis more fully into 
considerations of regulatory policy, and in focusing attention on objective 
price-output results. Employing it, together with more general theory, we 
may be able to develop economic ends for policy, although in an undulv 
simplified form if the new theory refers primarily to static equilibrium 
situations. Since the new theory is primarily a description of unregulated 
behavior, it does not suggest much about the means for attaining these 
ends via regulation. Criteria of the appropriate means of policy emerge 
from a wide range of considerations somewhat broader than those ordi- 
narily entertained in a 'priori economic analysis. 

Empirical studies of pricing and competition, which are formally less 
restricted and can be more specific about the setting of observed behavior, 
have made contributions to the critique of past policy and the formula- 
tion of future regulative methods, at least as applied to specific industries. 
As represented in the more detailed industry studies, research of this sort 
permits a quantitative measurement of price-output results, whereas a 
priori predictions would give only general and perhaps very imprecise 
guides. Normative evaluation of these results is of course still necessary, 
and the equipment for this evaluation is quite imperfect. Given depend- 
able evaluation, however, the detailed empirical studies may offer a reli- 
able guide to the occasions for interference. They also offer specific 
insight in specific situations into the practicable alternatives to observed 
behavior, and into the character of possible policy means of attaining 
them. De Chazeau's appraisal of the basing-point problem in steel is a 
good example of the usefulness of this sort of research in shaping policy, 
and there are other instances of similarly good work. The most evident 
limitations are the large number of detailed analyses required to furnish 
a basis for comprehensive public policy, and the serious lag in the devel- 
opment of adequate norms for evaluating behavior. Nevertheless, recent 
general critical studies of regulatory policy, including a number of re- 
interpretations of American policy toward competition and concentra- 
tion of industry, seem to reflect the beneficial influence of price policy 
studies. 81 

Special emphasis should be placed on two currents of policy thought 

81 Any general consideration of literature on regulation lies outside the intended scope 
of this paper. Illustrative of the trend mentioned, however, are Arthur R. Burns, "The 
Anti-Trust Laws and the Regulation of Competition," Law and Contemporary Problems. 
June 1937, IV, pp. 301-320; and Paul T. Homan, "Notes on the Anti-Trust Law 
Policy," Quarterly journal of Economics, November 1939, LIV, pp. 73-102. 


which have been engendered by empirical research into price policies: 
one stemming largely from the writings of E. G. Nourse, and the othei 
from those associated with the Arnold administration of the anti-trust 
laws. Both tend to modify in some degree the rather negativistic policy 
suggestions early drawn from Chamberlin's analysis. As suggested above, 
the work of Nourse and Arnold has been useful on a theoretical level in 
reminding us that precise objective determinacy of price via independ- 
ent action has not been demonstrated— that there is possibly a significant 
range of choice for the entrepreneur (Nourse), that market structures are 
not immutable, and that the price equilibria so precisely set forth in 
geometry may in practice not result from strictly independent action but 
may more often than not owe a great deal to good old-fashioned collusive 
activities (Arnold). 

The positive policy thesis of Nourse, based on his general analysis and 
on historical studies of price policies and their makers, is that industrial 
price policies may be influenced significantly— particularly in the direc- 
tion of "low" instead of "high" prices— by appropriate education and 
persuasion of price makers. This alteration of pricing is supposedly con- 
sistent with "maximum" profits, whatever those may be in a dynamic 
economy. This is not an idea to be rejected a priori, especially in an econ- 
omy where "everyone is an economist," where secondary as well as pri- 
mary effects of pricing actions are possibly taken into account, and where 
the "public relations" effect of pricing decisions may be viewed as signifi- 
cant. But the establishment in theory and in practice of a significant 
range of indeterminacy for the price-output decision does not guarantee 
the feasibility of Nourse's program. 

The Arnold proposal for policy involved a very vigorous enforcement 
of the anti-trust laws against collusion (and this was indeed a novelty) 
together with amendment or repeal of laws favoring collusion or restraint 
of entry into various industries. The success of such a program as a 
means of securing generally desirable price-output results is of course not 
a foregone conclusion. But neither theory nor empirical study has yet 
demonstrated that a regime of concentrated industry leads to given and 
unalterable price output results regardless of collusion, nor shown the 
extent to which eliminable collusive activities have been strategic to the 
observed pattern of price behavior. 

The present state of thought concerning public policy toward pricing 
and competition— if compared to that prevalent in 1930— serves to remind 
us of the subsequent accomplishments of price analysis and research in 
redefining problems and in increasing our general knowledge of the char- 


l /5 

acter of enterprise behavior. It also suggests the extremely inconclusive 
and fragmentary character of our solutions to these problems at present, 
and the nascent character of much of the needed research endeavor. A 
review of price policy research ten years hence may reveal the fruition Ca 
some of the early exploratory work. 


Arthur Smithies 

I. The Evolution of Fiscal Policy 

Under the impact of depression and war, the theory of the relation 
of fiscal policy to the working of the national economy has made great 
strides. While important details remain to be worked out and formula- 
tions can be made more elegant, the theory is well advanced. Only twenty 
years ago fiscal policy— a policy under which the government uses its 
expenditure and revenue programs to produce desirable effects and 
avoid undesirable effects on the national income, production, and em- 
ployment—was practically unknown. Except in times of war the policy 
of the United States during the nineteenth and early twentieth centuries 
was to hold government expenditures to a minimum, to pay the cost by 
taxation, and to retire the national debt as rapidly as feasible from the 
political point of view. This simple approach required very little in the 
way of a theory of government expenditures, while the theory of taxation 
was concerned primarily with the question of equity. Edgeworth, for in- 
stance, stated flatly that "the science of taxation comprises two subjects 
to which the character of pure theory may be ascribed : the laws of inci- 
dence, and the principle of equal sacrifice. >>:L 

One break with tradition had occurred in 1923 when the President's 
Conference on Unemployment recommended curtailment of public 
works in boom times and expansion in times of depression. While it was 
recognized that repercussions would spread to the production of materials 
for construction, there was no recognition of "multiplier effects" on con- 
sumers' demand. 2 

It was not until the Great Depression had descended on this country 
that the idea of a positive fiscal policy to promote recovery aroused real 
intellectual interest. It was only human suffering and political stress that 
compelled an economy-minded President to espouse it. In December 
1933, Keynes was able to say, "You, Mr. President, having cast off such 

1 Papers Relating to Political Economy (London, 1925), Vol. II, p. 64. 
8 Business Cycles and Unemployment, Report and Recommendations of a Committee 
of the President's Conference on Unemployment (New York, 1923), p. xxviii. 


fetters [of orthodox finance] are free to engage in the interests of peace 
and prosperity the technique which hitherto has only been allowed to 
serve the purposes of war and destruction." 3 Nevertheless, the theory of 
fiscal policy was to remain inchoate for some years and there is no sign 
that the President ever enjoyed his emancipation. 

The theory of those early years was mainly of the "pump-priming" or 
"shot in the arm" variety. Government expenditures w r ere expected to 
start the economy on its upward course. Recovery would permit emer- 
gency expenditures to be tapered off and higher revenue yields would 
produce a balanced budget. There was little thought of a systematic anti- 
cyclical policy, far less of the need to adapt fiscal policy to long-run eco- 
nomic objectives. 4 

The publication of the General Theory in 1936, the depression of 
1938, and the work in this country under the leadership of Professor 
Hansen brought great advances. In 1938 the President avow 7 ed for the 
first time that he was striving for recovery through the effects of his fiscal 
policy on the national income. By the end of the decade Hansen had 
made a convincing case for anti-cyclical policy and had propounded his 
famous thesis that fiscal policy was required to offset secular stagnation. 5 
Much can be said for and against the stagnation hypothesis and the argu- 
ment will only be settled by events. For our present purposes, the impor- 
tant point is that the need for a positive long-run fiscal policy had become 

The war gave a new impetus to thinking on fiscal policy— this time 
with the prevention of inflation as its main objective. Again Keynes was 
in the vanguard, 6 but was soon followed in the United States. 7 The Presi- 
dent's budget messages urged higher taxes than had ever been imposed. 
Proposals for compulsory savings were advanced, but foundered on ob- 
durate resistance within the administration. In the end, although taxes 
were raised higher than ever before, this country relied mainly on direct 
controls. This meant that we have as yet no answer to one main question 
of fiscal policy: is it possible to prevent inflation and achieve maximum 
production at the same time? 

8 Public letter to President Roosevelt, New York Times, December 31, 1933. 

4 It is difficult to document these statements. In this country most of the fiscal policy 
discussion took place within the government; the leading academic economists were 
neutral or hostile. The outstanding exposition of the pump-priming theory was Keynes' 
The Means to Prosperity (New York, 1933). 

6 See particularly his Fiscal Policy and Business Cycles (New York, 1941). See also 
Temporary National Economic Committee, Hearings, Part IX, for the views of many 
economists whom space prevents me from mentioning here. 

y How to Pay for the War (London, 1940). 

7 See A. G. Hart et al., Paying for Defense (Philadelphia, 1941); W. L. Crum et ai., 
Fiscal Planning for Total War (New York, 1942). 


In the thinking about postwar fiscal policy, the emphasis naturally 
changed from curing depression to keeping the full employment that had 
been achieved under the impact of war. It was plainly evident in 1941 
that full employment could be achieved through fiscal policy. Why could 
not the same thing be done, if necessary, in 1951 ? Consequently the idea 
of a government commitment to maintain full employment through fiscal 
policy became widely accepted. The British White Paper on Employ- 
ment presented the interesting spectacle of the new economics wrestling 
with Treasury tradition. The Canadian and Australian documents were 
more forthright in their advocacy of fiscal policy. The Employment Act 
of 1946 in this country originated as a proposal to achieve full employ- 
ment through fiscal policy alone. 

The second important feature of postwar thinking is the prominence 
given to taxation. After the wartime pressure for increased taxes to pre- 
vent inflation, it was natural that tax reduction should come into promi- 
nence as a device to increase production and employment. During the 
'thirties the major emphasis had been placed on higher expenditures, par- 
ticularly government investment, and, while expenditures were increased 
to promote employment, taxes were raised to appease tradition. Today 
there is much more thought of higher expenditures and lower taxes 
as alternative routes to the desired national income. Also, during the 
'thirties, it was widely held that steeply progressive taxes would not seri- 
ously retard production, since they would be paid largely out of income 
that would otherwise have been saved. Today there is a good deal more 
concern over the possible adverse effects of progressive taxation on the 
rate of investment and the rate of technical progress. 8 

Let me now turn to the major criticisms and qualifications that have 
been made to the argument for a positive fiscal policy: 

(1) A government commitment to keep full employment through 
fiscal policy would leave the government at the mercy of monopolists of 
business or labor. If a monopolist were to raise the price of his product, 
the government would be called upon to spend more money to avoid the 
unemployment resulting from the monopolist's action. Consequently, to 
avoid inflation a government must be equipped with adequate monopoly 
controls. This point is clearly made in the General Theory and is, I be- 
lieve, generally accepted. 

(2) In a full employment situation, the government must be able to 
alter its course rapidly if there is a change of signals. We cannot fly too 

8 It is interesting to compare Professor Hansen's views in Fiscal Policy and Business 
Cycles with his more recent opinions in Economic Policy and Pidl Employment (New 
York, 1947). 


close to the danger zone until we have better instruments and more 
knowledge. 9 Of course this criticism and the previous one are not argu- 
ments against a positive fiscal policy, but they are important qualifications 
on the extent to which it should be carried. 

(3) Reliance on fiscal policy perpetuates maladjustments and may 
obscure the need for economic reforms. For instance, public investment 
in times of depression may prevent reduction of construction costs which 
should come down if private construction is to revive. Again, fiscal policy 
may be used to offset the ill effects of monopolistic action and conse- 
quently remove pressure to go to the source of the trouble. 10 Should some 
unemployment be permitted while the basic adjustments are made? If 
they are not made, will there be a day of reckoning that cannot be 
avoided? Objections such as these pose dilemmas which the politician 
rather than the economist must resolve since they require comparisons of 
present gain with possible future loss. 

(4) Preoccupation with fiscal policy has diverted attention from other 
devices which should, along with fiscal policy, play an important part in 
a program for economic stability. In particular, it has been urged that 
monetary policy and debt management policy should be assigned more 
active roles. 11 This point of view has merit provided the scope of fiscal 
policy is not limited before it is known what contribution the other de- 
vices will make. 

(5) It is frequently argued that fiscal policy should not be used to 
achieve high levels of national income if it involves the national debt in- 
creasing faster than the national income. Such a policy is held to lead to 
socialism, communism, or fascism. Even if this were true, the possible 
alternatives, including mass unemployment, might prove to be shorter 
and faster routes to damnation. The argument usually stops before the 
implications of eschewing fiscal policy are explored. 

(6) Fiscal policy is objected to on the grounds that any departure from 
the rigid rule of annually balanced budgets would open the floodgates of 
government extravagance and should therefore be resisted. The National 
Association of Manufacturers, for instance, is uncompromisingly of this 
point of view. 1 " A most effective answer to this argument has been pre- 
sented by the Research Committee of the Committee for Economic De- 

! * Professor Albert Hart and other members of the C.E.D. research staff have done 
great service in emphasizing this point. See C.E.D. Research Staff, Jobs and Markets 
(New York, 1946). 

10 See the essays by J. M. Clark and Howard S. Ellis in Financing American Prosperity, 
Paul T. Homan and Fritz Machlup, eds. (New York, 1945). 

11 See C.E.D. Research Staff, op. cit. 

32 See The American Individual Enterprise System, The Economic Principles Commis- 
sion of the N.A.M. (New York, 1946). 


velopment. 13 It is pointed out that it the budget is to be balanced in 
depression, tax rates must be raised. In boom times, on the other hand, 
revenue is abundant, and if the budget is merely to be balanced there is 
ample room for extravagance. Thus, it is argued, both taxes and expen- 
ditures are likely to be higher under the balanced budget rule than if 
fiscal policy follows the rule of surpluses in boom times and deficits in 
depression. The C.E.D. is also rightly concerned with the question of 
extravagance, but attempts to combat it with reason rather than prejudice. 
Whether or not one agrees with the political philosophy of the N.A.M. or 
the C.E.D., one can readily agree that fiscal freedom cannot be permitted 
to undermine responsible government. One of the major purposes of this 
paper is to suggest ways in which freedom and responsibility can be 

Despite all qualifications and criticisms it is fair to say that, among 
academic and government economists, there is wide agreement that fiscal 
policy must form part of any program for economic stability. 14 There is 
also wide agreement that fiscal policy cannot be relied on exclusively. In 
particular, almost all economists would agree that it is essential to control 
or eliminate monopolistic practices. 15 

It is becoming more generally recognized that fiscal policy offers the 
best prospect of a conservative solution to the economic problem. Fiscal 
policy aims primarily at controlling aggregate demand and leaves to pri- 
vate enterprise its traditional field— the allocation of resources among 
alternative uses. It is therefore to be hoped that, through fiscal policy, eco- 
nomic objectives can be reached with less control by the State over the 
lives of its citizens than would be required by programs that called for 
the direct control of production and prices. The following quotation 
shows that at least one enlightened conservative is prepared to go some- 
what further than I would. ". . . in view of its responsibility to main- 
tain employment, the government must spend enough to close any gap 
between private (and business) spending and the total spending neces- 
sary to maintain full employment . . . However much the advocate of a 
liberal economy may hope for basic reforms which will eventually reduce 
the need for public spending, he cannot intelligently counsel economy 
and caution . . . Since reforms necessarily proceed slowly, private and 

13 Taxes and tJie Budget: A Program fur Prosperity in a Free Economy (New York, 


"See, for instance, A Program for Sustaiyung Employment (Washington, 1947), by 
the Committee on Economic Policy of the United States Chamber of Commerce. 

15 1 refer the reader again to F inancing American Prosperity. My statements here are 
confirmed by the measure of agreement that exists among Professors Clark, Ellis, Hansen, 
Slichter, and Williams. 


government outlays must at any time add up to full use of resources. Nat- 
urally this offers no defense of purposeless or wasteful use of public 
funds." 16 

II. Fiscal Policy and Budgetary Principles 

My own main criticism of the fiscal policy as worked out so far is that 
insufficient attention has been paid to its political and administrative as- 
pects. For instance, the "Functional Finance" 17 approach gives one the 
impression that the sole objective of the fiscal operations of the govern- 
ment is to work miracles on the national income. Discussions of pyramid 
building and of digging up bottles of bank notes provide graphic illustra- 
tions of the economic theory, but arouse reasonable doubts in the mind of 
the practical statesman. 

Unless the political issues can be satisfactorily resolved, fiscal policies 
cannot hope to succeed. I suspect that doubts on this question have fur- 
nished the basis for the Marxian theory of imperialism. In the opinion of 
neo-Marxians, it is politically feasible for a capitalistic country to embark 
on a course of foreign imperialism, but impracticable for it to expand 
investment within its own political boundaries. Tugan-Baronowski, in 
fact, agrees that pyramid building would do the trick, but he produces 
this example in order to ridicule the idea that adequate domestic policies 
will be undertaken, not to provide solace for the anti-Marxians. 

Fiscal policy cannot be worked out on the assumption that government 
expenditures or taxes can be justified solely by their effects on the na- 
tional income. If such a view became generally accepted, irresponsibility 
in government would be general. As Professor Schumpeter pointed out, 
Louis XV and A4adame de Pompadour were extraordinarily efficient 
spenders; yet their spending did not bring prosperity to France. To make 
fiscal policy work, economists must contaminate themselves with political 
theory and public administration. That is where budgeting comes in. 

The budgetary process in the United States is essentially one of pro- 
gram evaluation. As the functions of the government have grown more 
complicated, it has become essential to devise ways in which the relative 
merits of competing programs can be assessed, and in which the merits of 
the government's program as a whole can be compared with its cost. The 
result has been a budgetary process that is necessarily long and compli- 
cated, but has been increasingly successful in improving the efficiency of 

16 Howard S. Ellis, Of. cit., p. 137. 

17 See the celebrated article by Professor A. P. Lerner in Social ResearcJi, February 
1043, X, pp. 38-51. 


the government and in bringing essential issues before the President and 
the Congress for decision. 

The most notable step forward was the Budget and Accounting Act of 
1 92 1 . Before that act estimates were submitted by the individual depart- 
ments to the Congress with no review by the President and with no 
regard to the requirements of the budget as a whole. Moreover, the re- 
quests of each department were sent to one of the eight separate and 
independent appropriations committees of the Congress, so that in the 
legislative branch of the government also there was no unified exami- 
nation of the budget. 

The Act of 1921 requires the President to prepare a complete budget 
of estimated revenues and expenditures for the government as a whole; 
and thus the budget goes to the Congress as the recommended program of 
the President. Soon afterwards the Congress overhauled its own machin- 
cry for receiving the budget. The eight independent committees were 
replaced by one appropriations committee in each House. These two 
committees are now responsible for the entire budget in Congress. 

In 1946 a further important step was taken to improve congressional 
consideration of the budget as a whole. The Legislative Reorganization 
Act required that the appropriations and revenue committees of both 
Houses should meet jointly at the beginning of the session to prepare a 
"legislative budget" stating the objectives which the appropriations com- 
mittees should strive to attain for the budget as a whole. If this procedure 
proves workable, there will be machinery not only for examining the 
whole expenditure side of the budget but also for considering expendi- 
ture policy in relation to revenue policy. The experience with the legisla- 
tive budget in its first year has not been reassuring. Before the Joint 
Committee could agree on general targets the appropriations committees 
and the revenue committees had completed their work. The legislative 
budget never came out of conference. It is greatly to be hoped that the 
legislative budget procedure can be made to work, since, without it, Con- 
gress has no direct machinery for considering revenue and expenditure 
policies together. 

The Budget and Accounting Act created the Budget Bureau and made 
it directly responsible to the President. On behalf of the President the 
Bureau examines the requests of the agencies and translates into specific 
terms the President's budget policy. The Bureau is organized to assess not 
only the intrinsic merits of particular programs but, more importantly, to 
appraise their relative merits and to achieve a proper balance within the 
limits of the budget as a whole. 

In some quarters the Bureau has acquired an unenviable reputation oi 


always saying "no." Occasionally it does say "yes." During the war, for 
instance, there could be no question of hindering war production with 
financial limitations. In general, however, it is the proper function of the 
Bureau of the Budget to cut departmental requests. The departments are 
principally concerned with their own programs and have a natural tend- 
ency to extend and improve the services they render to the public. The 
Bureau on behalf of the President has to take into account the political 
and economic limits on the budget as a whole. It must compare expendi- 
tures with the receipts that can be expected. The departments would 
probably not be doing their duty if they did not request more than the 
President allowed, and the Bureau would not be doing its duty if it did 
not in general reduce the requests of the departments. 

The appropriations committees perform in the legislature somewhat 
the same functions as the Bureau of the Budget performs in the executive 
branch. It is their function to impose limits on the tendencies of subject- 
matter committees and of outside pressure groups to expand particular 
programs. It is also natural that the appropriations committees should 
have a somewhat more economical outlook on the budget than the Presi- 
dent. After all, the Congress rather than the President has to finance the 
budget. It would be an indication that forces in the legislature were not 
properly balanced if the appropriations committees did not have a repu- 
tation for economy. 

This brief description of the budgetary process indicates clearly that 
the requirements of good budgeting are by no means identical with the 
requirements for a positive fiscal policy. Of course the steps that have 
been taken to unify the budget are also indispensable for a coherent fiscal 
policy. But the process of program evaluation that is essential for good 
budgeting makes a flexible fiscal policy more difficult to achieve. It takes 
about six months to prepare the budget in the executive branch and about 
six months more for the Congress to consider it. Thus, budget plans must 
be begun about a year in advance of the time when they are to take 
effect. With our present limited foresight, fiscal policy may have to be 
adapted to changing economic conditions much more rapidly. How to do 
justice to both the program point of view and the fiscal policy point of 
view is one of the major problems yet to be faced. 

I may have given the impression that the Budget Bureau is interested 
only in economy. Anyone who reads the President's budget messages will 
discover that in formulating the budget he is by no means forgetful of the 
requirements of fiscal policy. Yet it remains true that the main preoccupa- 
tion of the Bureau is careful assessment and efficient execution of indi- 
vidual programs. 


The Employment Act of 1946 should bring the fiscal policy point of 
view more to the foreground in the deliberations in both the legislative 
and executive branches of the government. Under this act, the President 
is required to report to the Congress on all policies of government that 
will contribute to the achievement of maximum production, employ- 
ment, and purchasing power. The Congress is required to consider his 
report in a special joint committee. Fiscal policy certainly cannot escape 
consideration if the purposes of the act are to be achieved. The Council 
of Economic Advisers and the Bureau of the Budget must jointly work 
out for the President recommendations that will satisfy budgetary needs 
and also meet the requirements of fiscal policy. In the Congress, the fiscal 
views of the Joint Committee on the Economic Report must be fully 
taken into account in the recommendations of the appropriations commit- 
tees and the revenue committees. 

We now have the legislative and executive machinery required for 
budgetary and fiscal policy purposes. But machinery alone cannot do the 
job. There must be a much fuller understanding throughout the govern- 
ment of the national objectives. I have indicated that the proponents of 
fiscal policy have not fully appreciated the need for the rather tedious 
budget process, and that the budget process itself has been evolved with 
little regard for the need for a positive fiscal policy. Fiscal policy can be 
neither responsible nor acceptable without good budgeting. As was seen 
in the depths of the last depression, budgetary procedures can break 
down in the face of the very depressions that a proper fiscal policy seeks 
to avoid. 

III. The Objectives of Policy 

The function of the budget process 18 is to make a consistent whole of 
the entire policy of the government and to achieve balance among the 
various objectives of policy. Fiscal policy, on the other hand, is more 
directly related to the attainment of economic goals. 

The main objectives of policy, in my opinion, are national security, 
social security, economic and social progress, and political stability. That 
some of these compete with each other, while others complement each 
other, is a proposition that we can easily illustrate. 

18 1 say advisedly "budget process" and not Budget Bureau. As indicated above, the 
budget process continues until the budget is enacted by the Congress. In the executive 
branch, it is the responsibility of the President himself to carry out the budget process— a 
responsibility that he cannot delegate. The Bureau's function is to render the President 
assistance in^the process and to see to it that the major policy issues are brought to him 
for decision. 


It is sometimes thought that national security programs have a clear 
right-of-way in the budget. Nothing could be more erroneous. Any at- 
tempt to achieve complete security— if the term has any meaning— would 
require a budget very much larger than the present one. We are pre- 
pared to take national security risks for the sake of the other objectives of 

During the 'thirties, the preponderant view was that economic progress 
and social security went hand in hand, that every step toward greater 
social security also meant a more rapid rate of economic progress. An in- 
transigent minority held that every measure to remove economic hazards 
retarded progress. Now it is more widely recognized, on the one hand, 
that progress and security may clash and, on the other, that greater secu- 
rity through public means may avoid private attempts to obtain security 
through restrictive practices. 

Political stability, by which I mean no abrupt change in our political 
institutions or "the American way of life," is undoubtedly one of the 
dominant policy objectives of this country. Social and economic change 
is necessary for political stability, but too much change can destroy it. 
Both the reactionary and the radical are prepared to endanger political 
stability, the one by insistence on too little change, the other by insistence 
on too much. All the programs of government are subordinated to some 
extent to the objective of political stability. We are prepared to take risks 
with national security for the sake of political stability— a thoroughly ra- 
tional course when one of the main objectives of national defense is to 
protect our political institutions. 

These illustrations suffice to demonstrate the political character of the 
budgetary process. Various objectives of policy can be balanced and com- 
bined only through the interplay of political forces. In this paper, I am 
concerned mainly with the narrower field of economic objectives. Their 
attainment, however, must be considered as part of the budgetary process 
as a whole. And, as we shall see, the choice among economic objectives 
involves political decisions of the same character as the balancing of the 
economic with the non-economic. 

The main economic goals to which fiscal policy can be directed are as 
follows : 

( 1 ) In line with the tradition of economics, the first goal that should 
be mentioned is that of maximum economic well-being. Despite the pit- 
falls in the concept of economic welfare, I fail to see how we can get 
along without the idea. Nothing that I shall say will depend on the 
thesis that welfare can be measured in any absolute sense. My only 
assumption will be that we can generally decide whether it has increased 


or decreased. Of course, this is too general for many practical purposes, 
but we shall find it a useful starting point for our analysis. 19 

(2) A second goal, which has achieved wide national and interna- 
tional popularity, is "full employment," which means that all those 
willing and able to work should have an opportunity to work. It is still a 
matter of debate among its proponents whether "full employment" means 
that job vacancies should be more numerous or fewer than the number 
of the unemployed— in other words, whether labor should sell its services 
in a buyers' or a sellers' market. And if the latter, how much unemploy- 
ment should be counted on as a regular thing from the point of view of 
avoiding inflation and keeping competition in the labor market? 

As a general economic objective full employment is obviously incom- 
plete. No mention is made of the rate of real wages and no question is 
raised about the rate of progress of real incomes in the future. Little is 
said of the distribution of incomes or of standards of economic welfare. 
The idea is that if full employment is won, all other blessings will auto- 
matically follow. 

This approach can be attributed largely to the genius of Keynes for 
putting first things first. When the world was in a state of hopeless de- 
pression, it could easily be argued that if only employment could be in- 
creased by any means a democratic state could be persuaded to adopt, 
everything would be better. Total real wages, profits, saving and capital 
formation would all increase. The outlook for the future as well as the 
present would be improved. And who can contend that if the world had 
given its undivided attention to the attainment of full employment in the 
'thirties, the outcome might not have been better than it was? Neverthe- 
less, when we are trying to work out an approach to economic stability 
rather than to cure a deep depression, something more complex is re- 

(3) A third approach would be to substitute real income for employ- 
ment as a goal and to pay due regard to the effects of policy on the rate 
of accumulation of capital and the progress of technical knowledge. Tax 
policies in particular would be designed in part with a view to achieving 
an optimum, which is not necessarily a maximum, rate of capital accumu- 
lation. The effect of tax policies on the productivity of labor would be 
considered. Government expenditures would be distributed between de- 

19 I am fortified in this approach by finding that Professor Pigou uses it unhesitatingly; 
see his A Study of Public Finance (London, 1947). My own views, especially on long- 
term policy, have come mainly from practical experience with budgeting. I am gratified 
to find support for them in Pigou's book, which is strictly academic in origin. However, 
his discussion is limited in that he discusses the attainment of maximum satisfaction on 
the assumption that the operations of public finance do not affect the national income. 


velopment and welfare programs by a comparison of the present and 
future real income needs of the country. It may turn out that there is 
some conflict between a real income objective and a full employment ob- 
jective. A policy that maximizes employment of labor may not maximize 
real income and may not be consistent with the optimum rate of eco- 
nomic development. 

(4) It may be desired to stress not total production and employment, 
but the contribution made to the total by private enterprise. Privately 
produced goods and services are preferred to those produced by the gov- 
ernment, not because of any intrinsic superiority, but because there is 
political sentiment for leaving as much as possible of total production in 
private hands. To the extent that this objective is given weight, measures 
such as tax reduction and loans or guarantees to private enterprise would, 
on that account, be preferred to public welfare programs. 

(5) In contrast to (4), we have what I may call the planned welfare 
approach. The best example of this is Beveridge's Full Employment in a 
Free Society. He urges that the nation must exterminate the giant evils 
of want, disease, ignorance, and squalor. The need for a frontal attack on 
these evils is so great that the government must take whatever action is 
necessary, regardless of private enterprise or vested interests. If the gov- 
ernment takes its social obligations seriously, he argues, there will be no 
need to search for policies to increase production and employment. Bev- 
eridge believes that resources are so scarce that, to carry out these urgent 
social plans, government action must be extended to avoid waste. For 
instance, he believes that total investment should be subject to national 
planning. Beveridge is writing for postwar Britain. In the United States 
we can afford to indulge our whims more freely. 

(6) A further economic goal is greater equality of incomes— as an end 
in itself rather than as a means to increase total production or to raise 
living standards at the lower end of the scale. We can agree that the 
greater equality of incomes that comes from the elimination of restrictive 
practices will also increase production and employment. Also greater 
equality will reduce saving and increase effective consumer demand. But 
it is quite possible that egalitarian policies may reduce both business and 
labor incentives and so retard economic development. In Australia, for 
instance, the margins for skilled labor over common labor allowed by the 
compulsory arbitration system are very much narrower than the margins 
in this country. I suspect that this emphasis on equality, rather than on 
the incentives to become skilled, may have something to do with the 
lower productivity and its lower rate of increase in Australia as compared 
with the United States. Other examples come readily to mind. A country 


may have to choose between greater equality and greater productivity. 

(7) The last economic objective I shall mention is stability of income 
and employment. It is easy to agree that if the total income over a period 
is given, economic well-being is greater if income is stable than if it 
fluctuates. Periods of unemployment are not compensated by periods of 
overemployment. But to state the case that way begs the question. The 
real issue is : Will real income increase more rapidly if it fluctuates than if 
it does not? If so, is a more rapid rate of increase worth the price of alter- 
nate booms and depressions? 

In economic thinking about postwar policy, great stress has been put 
on the advantages of stability. With the benefit of hindsight, I am now 
inclined to think these advantages may have been overstressed. After 
making all allowances, countries that relaxed controls and allowed some 
inflation to occur seem to have increased production relatively more than 
those which did not. But the score can only be tallied after the "readjust- 
ment period," if any, has been undergone by the countries that experi- 
enced some inflation. Meanwhile, I do not think we should be too 
dogmatic in asserting the pre-eminent virtues of stability. 

I do not propose to try here to formulate the economic objectives of the 
United States, although I suspect they partake of each of the goals I have 
mentioned. For present purposes, I need only assume that there is some 
objective of national policy. In the Employment Act of 1946 the Con- 
gress, by a large majority of both parties, undertook to define our eco- 
nomic objectives. The act states: 

Sec. 2. The Congress hereby declares that it is the continuing policy and respon- 
sibility of the Federal Government to use all practicable means consistent with its 
needs and obligations and other essential considerations of national policy, with the 
assistance and cooperation of industry, agriculture, labor, and State and local gov- 
ernments, to coordinate and utilize all its plans, functions, and resources for the 
purpose of creating and maintaining, in a manner calculated to foster and promote 
free competitive enterprise and the general welfare, conditions under which there 
will be afforded useful employment opportunities, including self -employment, for 
those able, willing, and seeking to work, and to promote maximum employment, 
production, and purchasing power. 

The passage of the Employment Act may turn out to be one of the 
most important events in the legislative history of the country. Or, at the 
worst, it may prove to be nothing but an empty gesture. This act can be 
significant only if all branches of the government become thoroughly 
indoctrinated with it, and if Presidents and Congressmen can be over- 
thrown because they have failed to carry out its intent. 

Of course the declaration of policy must be discussed and interpreted. 


What precisely does "maximum employment, production, and purchas- 
ing power" mean? I believe it is one of the main functions of the Eco- 
nomic Report to clarify and explain the meaning of this act in the 
executive branch. It is one of the main functions of the Joint Committee 
to expose the objectives of the act to political debate and to convince the 
committees on special subjects that every program must be consistent 
with and, if possible, further the national policy objective. One of the 
great merits of the act is that it gives the President and the Congress an 
incentive to think on a plane of abstraction to which they have hitherto 
been unaccustomed. One can only hope that they will respond to this 

IV. The Mechanics of Fiscal Policy 

To reach any of the economic objectives we have described, it is neces- 
sary that a certain level of national income be attained. In this section we 
shall discuss the relation of government expenditure and taxation to na- 
tional income. We shall also see that a mere study of the mechanics of 
fiscal policy leaves undecided the question of what policy should in fact 
be pursued. 

Government expenditures may consist either of payments for goods 
and services or of "transfer payments/' such as veterans' benefits, unem- 
ployment compensation, or old-age pensions. 

To simplify the argument, I shall assume that the government deter- 
mines the total amount of tax to be collected, rather than the tax rates to 
be applied. 

Let us assume to begin with that the process of taxation and govern- 
ment expenditure does not redistribute income either among profits, 
wages and other factor incomes, or among personal incomes of various 

We shall also assume that the willingness of private enterprises and 
individuals to undertake expenditures depends on their incomes after 
taxes and also on circumstances independent of the government's policy. 


(1) An increase in government expenditure on goods and services 
with taxes unchanged will increase the national income by the value of 
the goods and services purchased by the government plus the induced 
effects on private consumption and investment, which I shall call the 
"repercussion effects." 

(2) An increase in transfer payments with taxes unchanged will in- 
crease the national income by the repercussion effects only. 


(3) A reduction in taxes with unchanged expenditures will produce 
the same result as an increase in transfer payments. 

(4) It follows that an increase in government expenditures or an 
equal reduction of taxes will have the same effect on private income after 
taxes and on private expenditures. 

(5) Equal increases of taxation and transfer payments will cancel each 
other out. 

(6) Equal increases of government expenditure on goods and services 
and of taxes will increase the national income by the value of the goods 
and services. The repercussion effects will cancel each other out. Conse- 
quently, private income after taxes and private expenditures will remain 
unchanged. 20 

E0 These propositions can be proved algebraically as follows : 

Let E = private expenditures on investment and consumption 
G — government expenditures on goods and services 
R — government transfer payments 
T — Taxes 
Y = national income 
It is reasonable to suppose that E depends partly on national income after deducting 
taxes and adding transfer payments, and partly on other factors, denoted by B, which 
are independent of present national income. Assume therefore: 

(i) E = a(Y-T + R) + B 

We have also: 

(2) Y = E + G 

Thus substituting for E in (2) 

(3) Y(l-a) = B + G-a(T-R) 


(4) AYr=AG + aAG (proposition 1) 

1 — a 
(T and R constant) 

(5) A Y = a A R (proposition 2) 

(G and T constant) 

(6) AY= — -a A T (proposition 3) 

(G and R constant) 

The term on the right of each of these expressions denotes the "repercussion effects." 

The increase in private income is A Y — AG. Thus if A G = A R = — A T, in (4), 
(5), and (6) respectively, the increases in private income and, from (1), the increases 
in private expenditure are equal (proposition 4). 

If T and R both increase and A T = A R, it is clear from (3) that Y is unchanged 
(proposition 5). 

If G and T both increase and A G = A T, we have from (3) 
A Y =_ A G (proposition 6) 

If the objective of fiscal policy is to achieve a given income goal, Y , we have from 


(7) G - a (T - R) = Y A (1 - a) - B 

It is obvious that an indefinite number of combinations of G, T, and R can be selected 
to give the required result. 


How far must these conclusions be modified when we consider the 
effects of fiscal policy on income distribution? There can be little question 
that the federal budget redistributes income in the direction of equality. 
The tax system as a whole is progressive. 21 But government expenditures 
are probably more equally distributed than all incomes. Government 
procurement and construction programs may be assumed to be distrib- 
uted among income classes in much the same way as the incomes ol 
the whole employed population. On the other hand veterans' benefit 
payments, unemployment compensation, old-age pensions, and welfare 
expenditures in general go mainly to the lower end of the income scale. 

During the past twenty-five years there have been great changes in 
the size and composition of the federal budget, and its redistributive 
effects have been considerable. Yet for all this, the relation of consumers' 
expenditure to disposable income seems to have been affected little, if at 
all. The statisticians, despite their disagreements on the subject, all seem 
prepared to relate aggregate consumption to aggregate income without 
taking into account changes in income distribution. While total saving 
varies greatly between incomes of various sizes, the saving out of incre- 
ments of income varies very much less— and that is the relevant factor 

So far as repercussions on consumption expenditure are concerned, 
therefore, there seems to be little to choose between general increase in ex- 
penditure and general reductions in taxes. Of course, it is easy to think 
of particular cases, such as a tax on the very rich to pay bounties to the 
very poor, where the effect is considerable; but so far there has not been 
enough political latitude for such devices to make them practically 

The redistributive effects of the budget may well have much more 
important consequences on private investment. Although it is difficult 
to verify statistically, profits after taxes have an important independent 
effect, I believe, on the rate of private investment. A general increase 
in taxation accompanied by a corresponding general increase in govern- 
ment expenditures, is likely to reduce profits and to slow down the rate 
of private investment. In other words, a tax reduction may be assumed 
to have more favorable repercussions on private investment than an equal 
increase in government expenditures. 

We assumed that private expenditures do not depend directly on gov- 
ernment expenditures. This may not be true. Free medical services would 
reduce private medical expenditures and this might increase saving 
rather than other kinds of consumers' expenditure. Old-age insurance 

21 See Helen Tarasov, Who Does Pay the Taxes? (New York, 1942). 


may tend to reduce saving. But the same statistical argument we used 
above indicates that these effects have not been important so far. 

Government investment expenditures may encourage or deter private 
investment. Regional development, for instance, opens up new fields 
for private enterprise, and the expansion of the automobile industry 
would not have been possible without the national highway program. 
On the other hand, the government can compete directly with private 
investment and so reduce its rate. 

Government expenditure programs may encourage private investment 
by providing a relatively assured market for the products of private enter- 
prise. I feel sure the aircraft industry considers the government a less 
volatile customer than the commercial air lines. 

All these factors must be taken into account when we attempt to 
assess the repercussions of an increase in government expenditures. 22 

We have discussed so far the effect of government expenditures and 
taxes on money national income and expenditures, while the prime ob- 
jective of fiscal policy is to achieve satisfactory rates of production and 
employment. By now all economists will agree that when resources are 
unemployed an increase in money national income will also increase 
real income. But there is also general agreement that the smaller the 
increase in money income required to evoke the desired increase in real 
income, the better. 

Expenditure and tax reduction policies may have different price effects. 
Expenditure programs can be and usually must be selective in their 
impact on industries or regions. The effects of tax reduction are more 
evenly spread over the economy. To restore production to a depressed 
industry or a depressed region through tax reduction may require so 
much tax reduction as to bring price inflation to the rest of the economy. 
On the other hand, to increase expenditures for goods and services may 
not be the most efficient way to cure a general slump. Production and 
prices might go up in some areas while others would remain depressed. 

Little has been said in this country of the possible effects of the budget 
on the productivity of labor. For a given real income, the higher govern- 
ment expenditures and taxes, the lower will be the probable rate of 
real wages— assuming that wage-earners do not count the services of 
government as part of their real wages. It is possible that high taxes lower 
willingness to work. This may not be important in the United States, 
but it is possible that heavy taxation in other countries has lowered the 
productivity of labor. 

I have tried to outline, in a very cursory way, the mechanics of fiscal 

" 2 See, for instance, Hansen, Fiscal Policy and Business Cycles, Ch. XII. 


policy. Despite the foregoing qualifications, the propositions we have 
stated above furnish a guide to the directions policy might take. I want 
to emphasize that, from the point of view of their influence on national 
income, taxation and expenditure adjustment must be regarded as alterna- 
tives. The choice of alternatives depends on what else the government 
wants to do besides achieve the desired income goal. 

It follows from what we have said that, within wide limits, a given 
national income can be reached in an indefinite number of ways. A 
study of fiscal mechanics alone cannot tell us what fiscal policy should 
be followed. The main purpose of this essay is to insist that the right 
fiscal policy cannot be determined until we also have a basis for deciding 
the relative merits of alternative programs. 

To simplify the discussion and to isolate the area of controversy, let 
us distinguish between long-run fiscal policy and short-run, or compen- 
satory, policy. From the long-run point of view, there can be little doubt 
that the government's taxing and spending operations must be planned 
with full regard for their economic impact. To what extent and in what 
manner short-run economic fluctuations should be compensated is more 

The distinction between long and short run corresponds to the distinc- 
tion that is made in modern business-cycle analysis between the equilib- 
rium value of national income (or any other central variable) and its 
actual value. 

The equilibrium value of income at any time depends on the growth 
of population, of technical knowledge, and of the stock of capital, and 
on the long-run propensities to invest and consume. It also depends on 
the programs and policies of the government. Public expenditure and 
taxation policies may hasten or hinder the course of economic develop- 
ment, and will affect the normal rate of employment of economic re- 

At any time the actual level of national income may and probably 
will diverge from its equilibrium value. The economy will be thrown 
out of equilibrium by external shocks such as war, famine, or revolution- 
ary changes in industry; and these shocks will result in alternate periods 
of prosperity and depression. The destruction and shortages of a war 
normally give rise to a postwar boom followed by a postwar depression. 
Here again the government's fiscal policy can exacerbate or offset fluctu- 
ations in private business. The normal tendency is for government to 
embark on new projects when income is high and revenue plentiful. 
Economic stability requires that this tendency be reversed. 

A long-run fiscal policy designed to keep the equilibrium national 


income at a satisfactorily high level would change little from year to 
year. The normal trends that should prevail in major expenditure pro- 
grams, such as public works and social welfare, and in taxation, would 
be known. As we shall see later, such a policy would require no radical 
departure from established tradition and would permit full justice to be 
done to the requirements of good budgeting. 

Legitimate controversy begins when compensatory action is considered. 
As we have already seen, some of the main objections to fiscal policy 
are related to the questions of whether the government should guarantee 
a certain level of employment and whether fiscal policy should cover up 
maladjustments. These objections raise doubts as to how far compensa- 
tory action should be taken. They are irrelevant to the question of long- 
run policy. 

V. Long-Run Fiscal Policy 

We have seen that an indefinite number of fiscal policies could be 
consistent with a given national income, and if national income were 
the only objective our problem would be indeterminate. As it is, it is 
necessary to find the fiscal policy that will help most effectively attain 
all or most of the country's objectives. That policy can only be found 
through the budget process of program evaluation. 

I can describe this process best by starting with a highly simplified 
situation. Suppose the sole function of government is to provide con- 
sumers' goods and services to the public and the sole purpose of taxation 
is to contract private spending in order to make way for public spending. 
For present purposes, transfer payments can be considered as negative 
taxes. I assume further that the objective of government is to adopt 
policies that will maximize well-being for the community as a whole 
and at the same time achieve the desired level of national income. 

Suppose the country is in a state in which its fiscal policy is prevent- 
ing it from attaining its national income goal and that new measures 
are being considered. Then, from what we have said in Part IV, it follows 
that a given increase in national income can be achieved either by increas- 
ing expenditure or by reducing taxation. If expenditures are increased the 
country gets more "government goods." If taxes are reduced, it gets more 
"private goods." 

If the social utility of additional government goods is greater than that 
of additional private goods, the appropriate policy will be to increase 
expenditures. Otherwise, taxes should be reduced. This process of 
comparing utilities should be carried on until the income goal is reached. 


If our starting point had been one of potential inflation, the alternatives 
would be to reduce expenditures, to increase taxation, or to make some 
combination of the two. The utility comparison again furnishes the 

Suppose the national income is at the prescribed level: the govern- 
ment's program may still be out of gear on utility grounds. Expenditure 
programs already under way may have a low utility while taxation is 
"oppressive," that is, its disutility is high. In that event expenditures 
and taxes should both be reduced until the marginal utility of expendi- 
ture is equal to the marginal disutility of taxation and the income ob- 
jective is reached with a lower budget. 

On the other hand, the right income might be attained, but the mar- 
ginal utility of the last increment of the prospective expenditure program 
might be greater than the marginal disutility of taxation. Then more 
expenditures and more taxes would be called for. Or the government 
could allow some price inflation to take place. In that event the con- 
sequences of the inflation should be judged in the same way as an in- 
crease in taxation. If the government is to obtain resources for its program, 
a levy must be made on someone.- 15 

To complete the picture, both the expenditure and the taxation side 
of the budget should conform to the utility rule. As ultimate objectives, 
the marginal utility of all expenditure programs should be equal and 
the marginal disutility of taxation should be the same for every taxed 

This approach lets the chips fall where they may so far as budget bal- 
ance and the national debt are concerned. If the need for a large govern- 
ment program were urgent— if , for instance, the whole nation had to go un- 
derground to avoid the atomic bomb— a budget surplus would probably be 
required if inflation were to be avoided by fiscal policy alone.- 4 If, on the 
other hand, the need for government goods were low and there were a 
strong preference for private goods, it might be necessary to budget for 
deficits to attain the desired level of income. If the national debt did 
not increase any faster than the national income, a deficit policy could 
be continued indefinitely without ill effects. If it did increase faster, 

23 Let us assume the existence of a welfare function (cf. P. A. Samuelson, Founda- 
tions of Economies'), W = W (Y, G, flT), which relates welfare to the national income, 
government expenditure on goods and services, and the goods and services aT, of which 
the private economy is deprived through taxation. This function increases with Y and G 
and decreases with T. The objective of the economy is then to maximize welfare subject 

• r -> v. -a v.- • ■ , 3W 3W 

to equation i^y) above, ror this it is necessary that — — — =1 — _ _. 

24 1 am not suggesting that such a policy should be pursued. As in the war, inflation 
would and probably should be controlled in part through direct controls. 


interest payments would occupy an increasing share of the expenditure 
side of the budget, or taxes would have to be raised to avoid inflation. It 
is interesting to note that, in a well-ordered economy, deficit financing 
should be associated with a limited view of the functions of government, 
while to finance a more "radical" program, surpluses may be required. 

If it is required that budget balance or some other budgetary condition 
should be the rule, either the income goal or the utility condition must 
be abandoned. If our analysis required a deficit to reach the income goal, 
the achievement of the same goal with a balanced budget would require 
expenditures of low social utility and taxes of high disutility. 

We have so far tacitly assumed that the money and the real income 
goals are fixed. It should not be overlooked that with continuing tech- 
nological progress the goal will increase from year to year. If prices are 
to be kept stable the money income goal and the real income goal will 
increase by the same proportion. With an increase in real income, the 
social disutility of a given amount of taxation will decrease. Thus, there 
would be room to consider the adoption of expenditure programs of 
lower priority. Depending on the usefulness of such programs, tax rates 
should be held the same or reduced. 

The utility of expenditures depends on political and social attitudes 
toward the function of government and the services it should perform. 
The long-run trend in this country appears to be in the direction of 
increasing the functions of government. This change in attitudes may 
preclude any downward tendency in the appropriate tax rates. 

The long-run policy would determine tax yields and expenditures at 
high levels of employment and income when no compensating action 
is required to offset deflationary or inflationary influences in the private 
sector of the economy. 

This policy would be subject only to gradual change. It might require 
a surplus or a deficit or balance in the budget every year. There is noth- 
ing but tradition that would make such a formula more difficult to apply 
than one that required budget balance every year. 

It will be seen that my solution for long-run policy corresponds for- 
mally with the "stabilizing budget" advocated by Mr. Beardsley Ruml and 
the C.E.D. But there is an important difference in substance. The C.E.D. 
maintains that the cash budget of the Federal Government should yield 
a surplus of $3 billion at high levels of income. It reaches this conclusion 
largely by deciding in advance the amount of debt reduction that should 
be undertaken. Expenditures and taxation must be consistent with that 
objective. In my scheme, expenditures and taxes are considered on their 
merits, and the behavior of the national debt is the resultant. Both 


schemes are equally feasible from the point of view of practical opera- 
tion, but I feel that mine puts first things first. 25 

We shall next try to bring the argument closer to reality by considering 
some of the major questions connected with the receipts and expenditure 
sides of the budget. Of course, in an article of this length, it is possible 
only to offer some discursive remarks on some of the major questions of 
public finance. 

VI. Long-Run Policy: Receipts 

(1) Taxation. The principle of equal marginal disutility that we 
have used above means that the tax system is designed to raise given 
amounts of revenue with the least possible sacrifice to the taxpayer. 
If the notion of maximum satisfaction or well-being has any validity, 
the principle of least sacrifice has much more validity than the principle 
of equal sacrifice. I suspect, however, that our present tax system 
corresponds more closely to the latter principle. 

The principle of least sacrifice, if fully applied, would presumably 
mean that all incomes of taxpayers would eventually be reduced to equal- 
ity, and that the marginal utility of the income left to each taxpayer 
would be lower than that of any non-taxpayer. If the principle of maxi- 
mum well-being were applied fully to the budget as a whole, it would 
be necessary to go further. Transfer payments would be made to non- 
taxpayers until the marginal utilities of all incomes were equal. Transfer 
payments very much of this type do occur in the family endowment 
schemes of Canada and Australia, but have won little support in the 
United States. In this country, transfer payments are usually made either 
to benefit special groups— e.g., veterans or unemployed— who have special 
claims, or to further specific social objectives— e.g., better nutrition. 

There are strong arguments why least sacrifice cannot be completely 
accepted even as an ultimate objective. For example: 

(a) There is a solid basis for the view that paying taxes makes for 
political responsibility, and consequently some taxes should be paid by 
the bulk of income recipients. 

(b) Experience with wartime tax systems has shown that high taxes 

25 It is beyond the scope of this paper to consider the question of autonomous wage 
pressure on the price level, hut I should indicate my opinion of the relation of this kind 
of pressure to fiscal policy. The possibilities of autonomous wage increases have a bearing 
on how much employment should be regarded as "full" and therefore on fiscal policy. 
But once a wage increase has occurred, fiscal policy should be adapted to the situation; it 
should not be used in an attempt to lower wages. Such an attempt could only lead to 
unemployment and loss of production. The main answer to the wage question must be 
found in the labor market itself. 


can lead to social waste and extravagance. It is doubtful whether tax 
enforcement could ever eradicate dubious expense items when the "gov- 
ernment pays 90 percent/' although this objection might apply less 
forcefully to the individual than to the corporation income tax. 

(c) The degree of income equality after taxes that least sacrifice im- 
plies might prove inconsistent with the desired rate of capital accumu- 
lation and economic progress. The principle of least sacrifice should be 
applied to both the present and the future. 

In general, tax systems based on relative degrees of sacrifice do not 
pay enough attention to the effect of taxation on incentives and pro- 
ductivity. Of course, manifest inequities can undoubtedly impair pro- 
ductivity, but there is no reason to believe that the tax system that 
achieves the greatest equity will also provide the most effective stimulus 
to productive effort. Unfortunately we have no empirical evidence that 
is in any way conclusive of the effects of taxation on incentives, and the 
prospect of getting any is not bright. I doubt whether time-series analysis 
will settle the question; and the method of direct inquiry obviously can 
only produce biased results. Tax policy will probably continue to be 
an arena for political as much as economic argument. 

We can conclude, however, that the principles of least and equal 
sacrifice provide the limits to an acceptable tax system. For the reasons 
indicated, it is not desirable to go as far in the direction of progressiveness 
as least sacrifice would require. On the other hand, it would probably be 
politically impossible to adopt a tax system that imposed less sacrifice 
on high incomes than on low incomes even though economic arguments 
were found for it. 

We have seen that all forms of taxation involve a transfer of private 
savings to the government. Are there any grounds for modifying our 
conclusions on the argument that taxation of saving diminishes the 
supply of funds for private investment? If the money market were ho- 
mogeneous, there would seem to be no validity in this argument. Assume, 
to begin with, that the taxation leaves the rate of private investment 
unaffected. Then it follows that the national debt will be reduced by 
the amount of the savings transferred below what it would have been 
had there been no transfer of savings to the government. This means that 
funds equal to the savings transferred will be available for private invest- 
ment. Thus our original assumption that the rate of investment is un- 
affected by the taxation can be retained so far as the supply of funds 
is concerned. To illustrate: Suppose the government plans to spend an 
extra $1 million. To avoid inflation it imposes new taxes of $1.2 million. 
As a result private consumption is cut by $1 million, offsetting the in- 


flationary effects of the new expenditures, and savings are reduced by 
$200,000. The government will use the $200,000 to buy government 
securities from private holders, who will then be able to increase their 
holdings of private securities by the same amount. Thus while the 
taxation diminishes the rate of increase of assets in private hands, it 
does not diminish the flow of new funds into private investment. 

With a non-homogeneous money market, our conclusion must be 
modified. It is possible that the taxation of saving may dry up sources 
of funds that normally go into risky ventures, while the funds released 
by the government seek a safe resting place. To the extent that this is 
true, our argument on utility grounds must be modified. The govern- 
ment would then have to consider whether a less equitable tax svstem 
and payment of services on a higher national debt were justified by the 
added stimulus to private investment. 

My last observation on taxation relates to Mr. Colin Clark's brilliantly 
suggestive generalization, based on inductive evidence, that no dcmo- 
cratic state will tolerate, on a permanent basis, taxes in excess of 25 per 
cent of its national income. 2,} In my terminology this means that the 
marginal disutility function of taxation becomes inelastic when taxes 
approach this limit. If inflation is to be avoided, a limit is thus placed on 
expenditures. Or, if expenditures are increased beyond that limit, in- 
flation must be met by direct controls. Direct controls may be regarded 
as imposing for a limited period less disutility than taxation, since, with 
them, the public is permitted to accumulate savings which compensate 
present deprivations with a claim on future production. Tax yields in 
the United States are today not far below Mr. Clark's limit, and I have 
seen nothing in our recent political history to prove that he is wrong. 
His generalization may turn out to be as hard to believe in and as hard 
to refute as the Paretian alpha. 27 

(2) Social Security-. Ever since the Federal Social Security System 
was inaugurated there has been keen controversy on how it should be 

(a) Should it be based on the contributory principle, that is, should 
it be financed chiefly by taxes which fall most heavily on the sector of 
the community that derives the greatest benefit? 

26 "Public Finance and Changes in the Value of Money," Economic journal, Decem- 
ber 1945, LV, p. 371. 

27 One of my critics has questioned my whole discussion of the disutility of taxation 
on the grounds that so little is known of the laws of incidence. I can only say in reply 
that, whatever the laws of incidence, taxes are borne by the economy as a whole, and 
political judgments are made as to whether the general level of taxation is too low or 
too high. I therefore believe that attempts to generalize about taxes as a whole are useful 


(b) Should reserves be accumulated now to avoid raising taxes in the 
future when old-age payments increase as the population grows older? 

In terms of our analysis so far, the answer to the first question is that 
the social security system should not be considered in isolation. It should 
be treated as part of the total government program and the extent to 
which it is contributory will depend on the general weighing of advan- 
tages and sacrifices. 

To the second question our analysis gives an unqualified negative. 
The accumulation of a cash reserve is in itself deflationary. To offset 
it, either government expenditures must be increased or other taxes 
must be lowered. The disbursement of the reserve would in itself be 
inflationary. To avoid the inflation taxes would then have to be increased 
or expenditures reduced. In fact, the budgetary situation would be the 
same as it would have been if the reserve had never been accumulated. 
Thus, there is no justification for distorting the budget in order to make 
possible the accumulation of the reserve in the first place. 28 

As far as economics goes, social security should be financed on a pay-as- 
you-go basis, and the contributory principle should not necessarily be 
adhered to. But there is more than economics involved. Opponents of 
social security feel they can limit its extent by keeping to the contributory 
idea. And many of its supporters believe that the program is rendered 
immune from political cuts and interference if the analogy with private 
insurance is maintained. They are even prepared to accept lower benefits 
in exchange for the greater political security and social dignity of the 
contributory system. In addition, there is still a vigorous reluctance in 
this country to support programs which allegedly provide something for 
nothing. Thus, while it is still open to debate how far the contributory 
principle will be carried, there is virtually unanimous agreement that it 
should not be abolished. 

(3) Sale of Government Services. While the government finances 
most of its program by taxation or borrowing, it habitually sells 
some services, such as postal services. Why should not the services of the 
post office be provided on the same basis as the services of the army or 
the police force? The answer is that the government undertakes to supply 
as much postal service as is required, and the demand for it is elastic. 
If it were provided "free" the demand would expand until it was used 
to satisfy needs of very low priority, and the program would not be 
justified on the basis of the tax criterion. The need to charge fees could 

28 This is from the long-run point of view. We shall see later that from the short-run 
viewpoint accumulation and decumulation of reserves furnish a useful element of built-in 


be avoided if the total supply could be controlled by the government 
and rationed among consumers. This is feasible with policemen, but 
would obviously create unwarranted inconvenience and administrative 
difficulty if applied to the post office. The general rule is that where 
demand is sufficiently elastic, fees must be charged. But there are " 
exceptions. It may be that the demand at zero price is not too large from 
a public policy point of view. The demand for policemen is probably 
clastic, but if fees for them were charged, we should almost certainly 
have too few police. Again, I feel that many government documents, 
now distributed free in the interests of general enlightenment, might 
be found to have very elastic demands if fees were charged. The demand 
for highway services is probably elastic, but the toll system was abolished 
—presumably on the grounds that the cost of free roads to the govern- 
ment was justified by the encouragement they gave to economic devel- 

If the demand is inelastic, whether or not fees should be charged 
should depend on who benefits from the services. If the services benefit 
all sections of the community, there is no objection in principle to pro- 
viding them free. If they are designed to benefit a particular group, fees 
as a rule should be charged. It is argued, for instance, that charges should 
be made for the use of the airways on the grounds that free airways give 
the airlines a competitive position in relation to land and water carriers 
that is unduly favorable. But this criterion should be applied with care. 
Any given recreational facility provided by the government will benefit 
most those who live closest to it, and that would suggest that fees should 
be charged. However, it may be reasonable to assume that one way or 
another, the government provides recreation equally for all parts of the 
economy. In that event there would be no objection to providing all rec- 
reational services gratis, assuming the demand is sufficiently inelastic. 

If fees are charged they should, I believe, cover the full costs of the 
services to the government. This is at variance with a prominent school 
of thought which holds that the rule should be marginal costs. 29 I have 
no satisfactory proof that full costs is the correct rule, but I do believe that 
the marginal cost principle would do injustice to the programs for which 
fees are not charged. The difference between average and marginal costs 
of the services must be financed through the budget, and would compete 
with other services. Can it be proved that to allow the postal deficit to 

20 See Harold Hotelling, "The General Welfare in Relation to Problems of Taxation 
and of Railway and Utility Rates," Econometrica, July 1938, XVI, pp. 242-269; and 
A. P. Lerner, The Economics of Control (New York, 1944). For an interesting alterna- 
tive solution, see R. H. Coase, "The Marginal Cost Controversy," Economica, August 
1946, New Series XIII, pp. 169-182. 


compete with the national defense program would be preferable to rais- 
ing postal rates sufficiently to eliminate the deficit? 

To charge fees based on cost does correspond generally with our tax 
criterion for other types of expenditure, but the need to operate on a fee 
basis prevents those operations from being financed in a way that will 
help to produce the desired level of national income. Instead, a criterion 
of annually balanced budgets must be retained for this part of the govern- 
ment's program. 

VII. Long-Run Policy: Expenditures 

( i ) General Program Evaluation. Our earlier assumption that the 
government acts as a kind of collective economic brain is of course a 
drastic simplification. Over two-thirds of the budget at the present time is 
directed to non-economic objectives. The economic objectives we have 
set forth may conflict with each other. Full employment may not mean 
maximum real income. Maximum production by private enterprise may 
not mean maximum satisfaction. The term marginal utility should be 
replaced by some more general term, such as marginal net advantage, to 
give the impression that the government is seeking to do justice to a com- 
plex of objectives. 

As I have said above, the budgetary process is essentially political. The 
economist has no method of deciding whether money should be spent 
for public health or public education, or, above all, how the needs of na- 
tional security should be balanced against those of social security. Nor 
can he balance the advantages of expenditures against the burdens of 
taxation. Nevertheless, economic analysis is essential if good political de- 
cisions are to be made. And in many cases economic analysis can be 
decisive. Economic analysis figures much more prominently in the politi- 
cal documents of today than in those of the 'twenties and the 'thirties. 

The process of deciding on the relative merits of programs necessarily 
begins in the executive branch of the government. It is not feasible or 
desirable for the President to attempt to leave all political decisions, espe- 
cially those requiring economic judgments, to the Congress. If construc- 
tive leadership is to be exercised by the government as a whole, the 
President must take the lead. His communications to the Congress must 
take the form of positive recommendations— on both general objectives 
and specific programs. Moreover, the executive branch is better organized 
than the legislative to consider the complicated interrelations among the 
parts of the government program. Our discussion has demonstrated the 
importance of the single executive budget. We can hope that, as the 


years pass, the Congress will devote an increasing amount of its atten- 
tion to the broader policy aspects of the budget and less to matters of 
administrative detail. 

To achieve the best division of labor between the executive and legis- 
lature, it is essential that the President submit his recommendations in 
a way that makes clear the decisions underlying them. This has been 
done to an increasing extent in the messages transmitting the budget. 
Last year a significant improvement was made when it became possible 
to discuss the program of the government in terms of its major functions. 

The question of classification of expenditures may seem a humdrum 
and trivial matter, but that is by no means correct. Traditionally, the 
budget has been prepared in terms of organization units and in many 
cases there is no close correspondence between organizations and func- 
tions. For instance, the conservation and development of natural re- 
sources is a function performed by the Interior, War, and Agriculture 
Departments. The general function of transportation comes under the 
Interstate Commerce Commission, the Maritime Commission, the Civil 
Aeronautics Board, the Civil Aeronautics Administration, the Federal 
Works Agency, the Post Office and other agencies. If the budgets of all 
these agencies were presented independently of each other, they would 
give little indication of the government's policy in any field. 

Now that the budget is classified under thirteen major functions and 
more detailed sub-functions, it becomes more feasible for the govern- 
ment's program to be presented in the way recommended in this paper. 
When the Congress comes to debate the budget in terms of functions, a 
great step forward will have been made. 

The success of the political process depends very largely on the Civil 
Service. In my judgment, the Civil Service should avoid, as far as practi- 
cable, making decisions which in their nature are political. It is its func- 
tion to crystallize political issues and to present them for decision at the 
political level. If the Civil Service is to enjoy the confidence of successive 
Presidents and Congresses, it must not encroach on their fields. On the 
other hand, it is difficult for the Congress or the President to perform 
their functions properly unless they do have confidence in the Civil 
Service. Otherwise too much of their attention will be diverted from the 
policy questions that are their main concern. 

I have said enough to indicate that good ideas are not enough to make 
the budgetary process work as it should. Effective organization is almost 
equally important. As we saw early in this paper, great improvements 
have been made, especially in the last decade, but in its ordinary opera- 
tions, with which we are here concerned, the government of this country 


rarely attains the sureness and decisiveness that it displays in times or 

(2) Public Investment. Suppose our economic objective calls for an 
increase of taxation to finance a new dam or a bridge. How can we de- 
cide whether the increase of real income that the investment will bring 
in the future will justify the present sacrifice by the taxpayers: 1 To esti- 
mate the returns from private investment in a limited field is notoriously 
difficult. In the case of T.V.A., for instance, one would have to fore- 
cast the future industrial history of a large region. Who in 1933 fore- 
cast the use of T.V.A. power in the war; and who in 1948 can forecast 
the effects of the use of atomic energy on the need for hydroelectric 
power? These things cannot be done, and yet no one could reasonablv 
suggest that we should not have a public investment program merely 
because we cannot estimate its benefits. 

One way toward a partial solution would be to consider first the invest- 
ment needs of the country, public and private. There appears to be some 
ascertainable relation between the rate of increase of real income and the 
rate of increase of the stock of real capital. By comparing future benefits 
with present costs, it may be possible to work out a national investment 
policy. It would then be necessary to decide how much of the total could 
be contributed by private investment. The balance would be what is re- 
quired from the public investment program. I am inclined to believe that 
an approach on these lines would give a better indication of the proper 
size of the public investment program as a whole than an estimate ob- 
tained by summing the costs of individual "meritorious" projects. If we 
had such a check on the total, rough and ready methods could be more 
readily accepted in distributing that total among claimants. 

I am not suggesting that future national income should be the sole 
criterion of public investment policy. It is a responsibility of the Federal 
Government to help to remove income disparities among regions. People 
in one region have a legitimate claim to protection against floods caused 
by the agricultural practices in another. Many other examples could be 

It has frequently been suggested that, because of their investment 
character, public works should be financed by borrowing rather than 
through taxation. Our analysis does not support this point of view. We 
have argued that every expenditure program should be preferable to a 
reduction of taxation or should justify an increase— as the case may be- 
since taxation is the device for diverting resources from private to public 
use. Further, it should be impossible to say whether any particular pro- 
gram is financed "out of taxation" or "out of borrowing"— since all pro- 


grams should satisfy the tax criterion. Consequently, attempts to separate 
the finance of public construction from that of current expenses is likely 
to impair the process of program evaluation. 30 

There may be, however, one valid element in the argument. Suppose 
our economic objective requires budget deficits. Then it is important that 
interest should not become an increasing proportion of the national in- 
come. This consideration may require more emphasis on programs that 
tend to increase national income, and this may commend public works. 
But is this necessarily the case? Is it not possible that better education 
may do more to increase future production than investments in steel and 
concrete? This is one of the instances where principles that are admit- 
tedly good for private business might lead the government astray. 

(3) Subsidies to Producebs. Our principles provide a frima facie 
case against subsidies to private production. In general, it is not justifi- 
able to restrict through taxation purchases of the things people want in 
order to induce them to buy something else at an artificially low price. 
Nevertheless there are important exceptions to this general rule. For ex- 
ample : 

Subsidies to private investment may be justified on the ground that its 
social net product is greater than its private net product. A railroad that 
opens up new territory increases the national income by more than it 
increases the income of the railroad. Thus, from the public point of view, 
we can justify investment that would be unprofitable to the railroad if it 
undertook the whole cost. 

It has long been recognized that, economic w 7 ell-being can be increased 
by subsidizing industries with decreasing costs induced by external econ- 
omies, on the assumption that when the infant has grown up it will be 
self-supporting. The argument has the same validity and the same limita- 
tions as the infant-industries argument for tariffs. 

Agricultural subsidies are a general practice in industrialized countries. 
It is the function of the agricultural sector to help replenish the popula- 
tions of the cities. Under laissez-faire conditions, this is brought about by 
agricultural incomes falling below industrial incomes. It is not consistent 
with the principles of economic welfare to allow this income disparity to 
persist. However, a subsidy program that achieved income parity could 

80 This section has been written from the point of view of the United States at the 
present time. To apply it to a country in its early stages of development further explana- 
tion is required. In the first place, borrowing from abroad need not satisfy our tax cri- 
terion except for the service charges on the borrowing. The foreign borrowing itself 
increases the supply of goods and services to the borrowing country. There is therefore 
no need to release them from private use by taxation. The same reasoning would apply 
to public investment of domestic funds that produced a rapid increase of final output. 


check this desirable population flow to the cities. The perpetual dilemma 
of agricultural policy is to encourage necessary readjustments, and at the 
same time do justice to the farmers. 

When compared with other expenditure programs, subsidies become 
more acceptable to the extent that the private enterprise objective is con- 
sidered important. For instance, in this country subsidies to private 
housing are eminently respectable. Public housing except for the lowest 
income groups is viewed with alarm. 

(4) Loans to Business. A loan by the government to business at a 
lower rate than the private money market will offer— or a government 
guarantee of a private loan— is in effect a subsidy to the extent that the 
cost of borrowing is lowered, and should be judged on the same basis as a 
subsidy. It is frequently argued that there is no need to apply the tax 
criterion to a loan, since it will eventually be repaid to the Treasury. This 
is incorrect according to our principles. If the income goal has been 
reached, new loan expenditures should be offset by increased taxation or 
reduced expenditures in other directions. The repayment of the loan, 
however, is in the nature of a tax on the earnings of the enterprise 
financed by the loan. During the repayment period, other forms of taxa- 
tion can be reduced correspondingly or government expenditures can be 
increased according to social needs. 

I have assumed in this argument that loans are made only to selected 
industries or businesses. If the government succeeds in making credit 
generally cheaper, we cannot say that there is a frima facie case against 
it. Whether it is good or bad will depend on whether or not it is desirable 
to increase the rate of capital formation. Here again, if the income goal 
has been reached, the expansionary effects of the cheaper credit should 
be offset with increased taxation. In policy discussions, it is usually taken 
as axiomatic that the objective of policy, in the long run at least, should 
always be to make credit cheaper. In certain circumstances, a case could 
be made for raising interest rates and lowering taxes. 31 One of the lux- 
uries of the present period of full employment is that it provides a respite 
from thinking only of how to cure a chronic depression. 

VIII. Compensatory Policy 

We now come to our last question. How is long-run policy to be ad- 
justed from time to time to offset fluctuations in private economic activ- 

31 1 would remind the reader that I am here considering long-run policy, not counter- 
cyclical policy. There too a rise in interest rates might be regarded as an alternative to a 
tax increase; but it is improbable that an increase of interest rates and a tax reduction 
would be desirable at the same time. 


ity? My discussion here will be brief, since the subject has been treated 
at length elsewhere in this volume. 

One simple answer that has obvious appeal is that the entire long-run 
policy should be adjusted. In the event of a depression the entire expendi- 
ture side of the budget should be speeded up and taxes reduced. In short, 
the same principle that we have worked out for the long-run should also 
be applied to the short-run. Instead of making plans that will remain 
fairly stable from year to year, the budget should be made to fit the eco- 
nomic needs of each particular year. 

Such a policy commends itself from the point of view of economic 
theory, but it could not be carried out without serious impairment of the 
budgetary process. As we have seen, that process is necessarily lengthv, 
and programs may have to be adjusted quickly to meet changing eco- 
nomic conditions. It is preferable therefore to consider short-run policy as 
a supplement to long-run policy, and it should be designed to interrupt 
as little as possible the application of budgetary principles to the long-run 

We can consider compensatory devices according to whether they are 
CO "built-in," (2) administrative, or (3) legislative. 

(1) "Built-in" Devices. Built-in flexibility of the budget is achieved 
when certain programs fluctuate by law in a way that will offset private 
economic fluctuations. If our long-run policy is consistently carried out, 
the behavior of the budget will automatically tend to offset ups and 
downs in the private sector. 

With a progressive income tax, for instance, a larger proportion of the 
national income is taxed when national income is high than when it is 
low. It therefore operates automatically to check booms and depressions. 
The social security system accumulates funds at high levels of income 
when tax receipts exceed benefits, and it decumulates at low levels when 
tax receipts tall off and claims for benefits rise. The agricultural price- 
support program automatically expands and contracts as farm prices rise 
and fall. 

These examples indicate that there is already some built-in flexibility 
in the federal budget. Any scheme for extending Social Security benefits 
will bring more such flexibility. It may become feasible to tie tax rates to 
some economic indicator, but at present we do not have sufficiently pre- 
cise knowledge of economic interrelationships to make this feasible. 

Built-in devices commend themselves highly on budgetary grounds. 
Their automatic operation means that they come into operation without 
the need for hasty political or administrative decisions. Long-run policy 
can be preserved intact. 


However, these arrangements cannot be relied on exclusively if we 
want full compensatory action. For a considerable amount of inflation, or 
unemployment, as the case may be, must take place before they come into 
effect with any force. They can mitigate but cannot prevent inflation or 

(2) Administrative Devices. It has often been suggested that the 
executive should be granted discretionary power by the Congress to vary 
the rate at which programs are carried out or the rates at which taxes are 

The Congress does habitually make appropriations for more than one 
year for long-run construction or procurement programs, and the execu- 
tive does have the legal authority to control rates of expenditure from 
year to year. But these programs prove in practice exceedingly difficult to 
adapt to changing economic conditions. In many cases it takes one or 
two years for programs to become fully reflected in construction activity. 
Once a program has been enlarged, it is difficult to contract it without 
waste, to say nothing of resolute opposition from all interested parties. 

In the past, government corporations, such as the Reconstruction Fi- 
nance Corporation, have provided the administration with a flexible in- 
strument of policy. However, by the Corporation Control Act of 1945, 
the Congress asserted its determination to exercise greater control over 
government corporations; and the tendency now is to bring the corpora- 
tions under stricter budgetary control rather than to exempt them from it. 
Here there is a conflict between the long- and short-run points of view. 
From the long-run point of view, there can be no question that the pro- 
grams of the corporations should be subject to regular Congressional ap- 
proval; but if that is done the executive is deprived of a useful counter- 
cyclical weapon. 

It seems very doubtful that Congress would agree to give the Presi- 
dent discretionary power to alter tax rates. It also seems to me doubtful 
that the President, from a political point of view, would want it. 

On the whole, therefore, I believe that the practical possibilities of ad- 
ministrative devices are strictly limited. 

(3) Legislative Devices. I conclude that apart from built-in flexibil- 
ity, we must rely for compensatory action on measures that are passed by 
the Congress. In fact that seems to be in line with the intention of the 
Employment Act, which makes the President and the Congress partners 
in their responsibility for economic stability. 

Anti-depression action will, I believe, require special requests by the 
President for appropriations to carry out special projects such as roads or 
housing, or to provide relief in the case of distress. It should also require 


special action to reduce taxes on a temporary basis. This seems all very 
similar— painfully similar, some will say— to what was done in the 
'thirties. But there can be a difference. The 'thirties were a highly experi- 
mental period. Many things that were tried had to be rejected. Some that 
were retained were badly executed, and some were misdirected. The dif- 
ference can be achieved by advance preparation. We shall probably never 
have a blueprint on the statute book. But it is possible that the Congress 
could pass a tax law in advance— to be put into force by a joint resolution. 
And the Executive can, and, under the Employment Act, presumably 
will, get its plans ready. So long as the imaginative vigor of the 'thirties 
is not lost, a much better job can be done next time we are confronted 
with depression. 

What should be the objective of compensatory policy? Should it be to 
iron out all fluctuations, or should it allow mild ups and downs in busi- 
ness? Our discussion in the last paragraph has answered the question to 
some extent. If Congressional action must be relied upon, it is doubtful 
that action will be taken before some signs of depression have appeared. 
And in the present state of our forecasting knowledge, it is doubtful 
whether it should. 

Even if forecasting were good and action could be taken in time, it is 
not certain that complete stability should be the objective. It still remains 
true that trouble can be caused by price and wage maladjustments, 
which, if they are not to be perpetuated, should be allowed to correct 
themselves. There will continue to be shifts between industries and re- 
gions. It should not be national policy to prevent these changes, but they 
do involve depressions in some segments of the economy. 

There is no reason, however, why these changes which should be 
allowed to occur should have secondary depressing effects. There are no 
grounds for the belief that a general depression does anyone any good. 
The best way to avoid secondary effects is to reduce tax rates. I conclude, 
therefore, that tax reduction should always be part of anti-depression 

One further point: foreign investment policies should not be under- 
taken as compensatory devices. It is essential from the point of view of 
good international relations, that lending policy be designed to help cany 
out foreign policy and not domestic policy. There is the strongest political 
temptation to export unemployment. But in the interests both of domestic 
stability and international harmony, it should be resisted. 

It is much easier to propose a feasible legislative program of fiscal policv 
to offset deflation than to propose one to prevent inflation. While strong 
political forces can be mustered to relieve a depression, innumerable ob- 


stacles and pressures appear when any attempt is made to cut inflation. 
My conclusion therefore is that fiscal control of inflation must depend on 
prompt abandonment of anti-depression measures and the anti-inflation- 
ary safeguards that are built into the long-run program. 

In the field of compensatory action, I believe fiscal policy must shoul- 
der most of the load. Its chief rival, monetary policy, seems to be disquali- 
fied on institutional grounds. This country appears to be committed to 
something like the present low level of interest rates on a long-term basis. 
There is not much room for reductions to alleviate depressions, and it 
seems generally agreed that, with the national debt at its present size, 
any appreciable increases in rates would cause serious financial disorders. 
No one has been prepared to suggest an increase in long-term rates to 
check the present inflation. 

Other methods of adjustment, such as anti-monopoly policy, belong 
clearly to long-run rather than to compensatory policy. While many econ- 
omists have recommended elimination of monopolies in times of depres- 
sion, few, if any, have been willing to urge an increase in monopoly to 
check a boom. 

I have some hopes that in the area of administered prices big business 
may come to practice private compensatory fiscal policies. The practice of 
accumulating reserves in prosperous times and disbursing them as divi- 
dends when current profits are low is all in the right direction. A stable 
dividend policy could be supplemented and strengthened by a fluctuating 
price policy. It can be argued that the present inflation of food prices 
would be less if managed industrial prices were higher, and that these 
higher prices would be justified if there were assurance of sharp price 
cuts in the event of a depression. Attempts by government to enlist the 
aid of business management have not been successful in the past. They 
are not likely to be in the future so long as government retains its ambiva- 
lent attitude toward the question of monopoly control. 

Although this discussion has not been optimistic, the least that can be 
hoped for is that more government action will be taken to mitigate de- 
pressions in the future than was taken in the past. Built-in flexibility is 
now more potent; it is unlikely that tax rates will ever again be raised 
with the onset of depression; and the political demands for government 
action will be more insistent. Finally, the long-run budget will be a much 
higher proportion of the national income than it was before the war. 
With a budget of between $30 billion and $40 billion, it would be im- 
possible for the national income to fall to anything comparable with that 
of 1932 or 1933. 

The type of policy suggested here would require the use of an ex- 


traordinary budget in addition to the regular budget. In my view, the 
regular budget would represent the long-run program on the expenditure 
sides. It would, however, reflect the effects of built-in flexibility. Of 
course if the total program were designed to achieve full employment in 
the ensuing year, there would be no such effects. The extraordinary 
budget, on the other hand, would show the effects of the special revenue 
and expenditure measures proposed by the President. Depending on the 
time when action was required, it might not even be submitted at the 
same time as the regular budget. 

I know that extraordinary budgets are anathema to many authorities 
whose objectives are the same as mine— to improve the budgetary process. 
But I can see no other way to do justice to the requirements of good 
budgeting and an effective and well-timed fiscal policy. 32 The use of an 
extraordinary budget would permit separation of that part of the budget 
which can be the subject of long-run planning from the programs that 
must be adopted in response to immediate needs. With the extraordinary 
budget, compensatory measures could be dropped more easily after the 
need for them had passed than if they had become part of the regular 

The economist will doubtless be irritated with the intrusion of public 
administration into this paper, and the public administrator, if he reads 
it, with the economic jargon. The political theorists will consider it naive. 
However, someone must try to bring together the fruits of political, eco- 
nomic, and administrative theory it we are to have a successful fiscal 
policy. 33 

32 1 am happy to find that Mr. Gerhard Colm came to the same conclusion. See his 
article, "Comment on Extraordinary Budgets," Social Research, May 1938, V, pp. 

33 After I had exhausted all the space available for this paper, I found there were many 
things left unsaid or unnoticed. Despite the advice of my helpful critics, there is still no 
discussion of national debt management. Against my own inclinations, I have had no 
space to consider the international aspects of domestic fiscal policies. I have not been 
able to give enough attention to the other measures which should complement fiscal 

The paper may appear to be more an exposition of my own ideas than a review of 
economic thought. To that I would reply that it is a review of the spoken as well as the 
written word — much of what I have written is an attempt to systematize what has been 
discussed in government circles, and especially in the Bureau of the Budget, over a num- 
ber of years. 


Lloyd A. Metzler 

I. Introduction 

The interwar period was a period of extraordinary and perhaps unprece- 
dented developments in the field of international trade. The gold stand- 
ard, which had been abandoned during the First World War, was never 
re-established on a firm basis, and exchange rates of many countries un- 
derwent substantial fluctuations. The Great Depression of the 'thirties, 
which sharply reduced the level of output and employment in many 
countries, had a drastic effect upon international trade. Neither the 
timing nor the severity of the depression was uniform as between coun- 
tries, and the consequence was a serious lack of balance in the interna- 
tional payments and receipts of many countries. Partly to offset these 
discrepancies, and partly to guard the dwindling markets for goods, coun- 
try after country imposed additional barriers to international commerce. 
Even after the general economic recovery of the late 'thirties had 
begun, the trade restrictions and special trading arrangements were for 
the most part retained. The Second World War was of course the occa- 
sion for much more comprehensive and complete governmental controls 
of international trade, but even if the war had not occurred we should 
have inherited an enormously complex system of trade regulations as our 
legacy from the unstable 'thirties. During the war years, it became in- 
creasingly apparent that such a complicated system would not automati- 
cally revert to a system of unregulated multilateral trade, and that any 
attempt to restore the old system would require constant international 
supervision and co-operation. This was the genesis of such organizations 
as the International Monetary Fund, the International Bank for Recon- 
struction and Development, and the proposed International Trade Or- 
ganization. Whether these agencies will succeed in their avowed purpose 
of assisting in the establishment of a stable and relatively free interna- 
tional economic order remains to be seen. The difficulties encountered in 
postwar recovery and the general trend toward state intervention in eco- 
nomic activity have seriously complicated the problems of the new organ- 



In any event, it is not the purpose of this paper to discuss the future 
prospects for international trade or the future development of commercial 
policy. Nor do I intend to describe further the interesting history of inter- 
national trade during the interwar years. This empirical enquiry has 
already been performed in a number of admirable studies, and it seems 
neither necessary nor useful to summarize the work of these able econo- 
mists. 1 This paper is primarily a review or a summary of recent changes 
in the theory of international trade. The empirical developments are 
mentioned largely because of the profound influence which they have 
had upon economic theory. If the interwar period was a period of dis- 
turbed conditions in international trade, it was also a period of rapid 
change in the theory of international economics. The changes were so 
numerous, in fact, that it is quite impossible to summarize all of them. 
Nevertheless, it seems to me that the most significant of the recent devel- 
opments can be classified under four main heads as follows: (i) the 
balance of payments and the theory of employment; (2) fluctuating ex- 
change rates; (3) price theory and international trade; (4) commercial 
policy and the theory of international trade. Although the discussion 
below is by no means exhaustive, an attempt has been made to evaluate 
the principal contributions in each of these four branches of international 

II. The Theory of Employment and the Balance 
of Payments 

The revolution in economic theory which occurred in the nineteen- 
thirties had a profound influence upon almost all branches of economics, 
and this was no less true of international trade than of other specialized 
fields. Since the new approach to economics was primarily a reconsidera- 
tion of traditional ideas regarding money, interest rates, and prices, it was 
natural that the most important changes in international economics 
should have been in the monetary aspects of the subject. The revolution 
actually extended considerably beyond the monetary theory, however, as 
a later discussion of commercial policy will show. 

Prior to the publication of Keynes' General Theory, the monetary 
theory of international trade had been one of the most widely accepted of 
economic doctrines. For more than a century and a half, English econo- 

1 Perhaps the most interesting of these empirical studies are Seymour E. Harris, 
Exchange Depreciation (Cambridge, Mass., 1936); Margaret S. Gordon, Barriers to 
World Trade (New York, 1941); Howard S. Ellis, Exchange Control in Central Europe 
(Cambridge, Mass., 1941); U.S. Dept. of Commerce, The United States in the World 
Economy (Washington, 1942); League of Nations, International Currency Experience: 
Lessons of the lnter-war Period (Princeton, 1944). 


mists and others in the English tradition had believed that the monetary 
system operates in such a way that a country's balance of payments tends 
automatically toward a state of equilibrium. If one country had a deficit 
in its balance of payments with another, for example, it was recognized 
that part of its payments abroad would have to be made in gold, and it 
was believed that the gold movement would bring about certain price 
changes which eventually would restore an even balance of payments. 
As a result of the increased supply of money in the surplus country, and 
the reduced supply in the deficit country, prices and costs would rise in 
the former and fall in the latter. The deficit country would then become 
a relatively cheap market in which to buy goods, and its exports would 
rise while its imports declined. This process would continue, according 
to the classical view, until a balance between payments and receipts was 
again established. 2 The classical explanation of the balancing process was 
eventually modified to consider the influence of interest rates on capital 
movements, to allow for a fractional reserve banking system, to recognize 
the similarity between gold movements and changes in foreign balances, 
and in other respects as well, but in substance the theory remained essen- 
tially as it was originally developed by the early English economists. 

The important feature of the classical mechanism, for the purpose of 
the present review, is the central role which it attributes to the monetary 
system. The classical theory contains an explicit acceptance of the Quan- 
tity Theory of Money as well as an implied assumption that output and 
employment are unaffected by international monetary disturbances. 3 In 
other words, the classical doctrine assumes that an increase or decrease in 
the quantity of money leads to an increase or decrease in the aggregate 
money demand for goods and services, and that a change in money de- 
mand affects prices and costs rather than output and employment. The 
Keynesian revolution cast doubt upon both of these crucial assumptions. 
Say's Law of Markets, which had been the bulwark of both the Quan- 
tity Theory of Money and the classical theory of the balance of payments, 
was rejected, and the possibility of general overproduction or general un 
employment was finally acknowledged. In the course of this revolution, 
the monetary system, regarded as a director of economic activity, was 
relegated to a somewhat secondary position, and economists increasingly 
emphasized the effects of saving and spending habits upon the circular 
flow of income. 

After the foundations of the classical theory had crumbled, it was only 

2 See, e.g., John Stuart Mill, Principles of Political Economy, Ashley ed. (London, 
1909), Ch. XXI, Sec. 4. 

3 Cf . James W. Angell, The Theory of International Prices CCambridge, Mass., 1926). 
Ch. III-VI. 


a short time until a new explanation of the balancing process in interna- 
tional trade emerged. Although the new theory of the balance of pay- 
ments was a direct outgrowth of the General Theory, Keynes himself had 
little to do with it; the first contributions were made by Mrs. Robinson 4 
and R. F. Harrod. 5 Some of the practical as well as the theoretical impli- 
cations of the new doctrine were later investigated by Haberler, 6 Salant, 7 
Kindleberger, 8 Metzler, 9 Machlup, 10 and others. The essence of the new 
theory is that an external event which increases a country's exports will 
also increase imports even without yrice changes, since the change in 
exports affects the level of output and hence the demand for all goods. In 
other words, movements of output and employment play much the same 
role in the new doctrine that price movements played in the old. Before 
discussing the relation of employment to the balance of payments in 
detail, however, a brief account should be given of a number of empirical 
studies of the adjustment process which were published during the inter- 
war years, for these studies, although carried out along classical lines, had 
a profound effect upon later developments in international trade theory. 


At the suggestion of Taussig, several economists made detailed inves- 
tigations of the balancing process under conditions of both fixed and fluc- 
tuating exchange rates. 11 In each of these studies a period of time was 
selected in which a particular country's balance of payments had been 
subjected to a disturbing influence, and the manner in which the balance 
of payments had adjusted itself to this disturbance was then examined. 
The general conclusion of most of the empirical investigations was that 
the balancing process had occurred largely as envisaged in the classical 
theory; i.e., the price movements and gold movements had agreed with 
classical expectations. Taussig himself later made additional studies 

4 Joan Robinson, Essays in the Theory of Employment (New York, 1937), Part III, 
Ch. I. 

6 R. F. Harrod, International Economics, revised ed. (London, 1939), Ch. V. 
G Gottfried Haberler, Prosperity and Depression (Geneva, 1940), Ch. XII. 

7 William A. Salant, "Foreign Trade Policy in the Business Cycle," Public Policy, 
Vol. II (Cambridge, Mass., 1941), pp. 208-231. 

8 Charles P. Kindleberger, "International Monetary Stabilization," in Postwar Eco- 
nomic Problems, S. E. Harris, ed. (New York, 1943), pp. 375-395. 

Lloyd A. Metzler, "Underemployment Equilibrium in International Trade," Econo- 
metrica, April 1942, X, pp. 97-112. 

10 Fritz Machlup, International Trade and the National Income Midtiplier (Philadel- 
phia, 1943), passim. 

11 The best known of these studies are: J. H. Williams, Argentine International Trade 
under Inconvertible Paper Money, 1880-1 goo (Cambridge, Mass., 1920); Jacob 
Viner, Canada's Balance of International Indebtedness, 1900-19 13 (Cambridge, Mass., 
1924); Harry D. White, The French International Accounts, 1880-1 g 13 (Cambridge, 
Mass., I933)- 


which gave further support to this view. 12 At the same time, however, 
some of the evidence which Taussig accumulated led him to doubt the 
adequacy of the classical theory. It was not that the balance of trade and 
the level of prices failed to conform to disturbing influences in the man- 
ner envisaged by the classical theory. Quite the contrary, they appeared 
to conform too well and too quickly. When Great Britain increased her 
capital exports, for example, Taussig observed that the British balance on 
current account adjusted itself with amazing rapidity to the new capital 
position, even though both gold movements and changes in prices ap- 
peared to be relatively small. 

. . . The actual merchandise movements seem to have been adjusted to the shift- 
ing balance of payments with surprising exactness and speed. The process which 
our theory contemplates— the initial flow of specie when there is a burst of loans; 
the fall in prices in the lending country, rise in the borrowing country, the eventual 
increased movement of merchandise out of one and into the other— all this can 
hardly be expected to take place smoothly and quickly. Yet no signs of disturbance 
are to be observed such as the theoretic analysis previses . . , 13 

The smoothness and speed with which many countries' balances of 
payments seemed to adapt themselves to changing circumstances in the 
years before the First World War led Taussig to surmise that the classical 
theory might be an incomplete explanation of the adjusting mechanism. 
"It must be confessed/' he said, "that here we have phenomena not fully 
understood. In part our information is insufficient; in part our under- 
standing of other connected topics is also inadequate." 14 Even before the 
theory of employment was developed, historical studies thus indicated 
that the balancing of international payments and receipts might be at- 
tributable to economic forces not considered in the classical theory. De 
spite his misgivings, Taussig never abandoned the classical theory, for he 
could find no other explanation of the balancing process. 

Meanwhile, other empirical studies were being made along entirely 
different lines, and these cast further doubt on the effectiveness of the 
price adjustments envisaged by the classical doctrine. The interwar period 
was a period in which extensive studies were made of the elasticity of de- 
mand for individual products, and the studies showed, almost without ex- 
ception, that quantities sold were much less responsive to changes in prices 
than had formerly been suspected. 15 The elasticities proved in most cases 

12 F. W. Taussig, International Trade (New York, 1928), Ch. XX-XXV. 

13 Ibid., p. 239. 
11 Ibid. 

15 The pioneer work in this field was of course that of H. L. Moore and Henry Schultz. 
See particularly the latter's book, The Theory and Measurement of Demand (Chicago, 


to be less than unity, and in some instances they were so small as to be 
almost negligible. Studies of demand elasticities for imports as a whole 
were not made until a later date, but when they were made they con- 
firmed the supposition which the earlier studies of individual commodi- 
ties had raised that the physical volume of imports might not be respon- 
sive to changes in prices. Hinshaw, 16 for example, estimated an elasticity 
of demand for imports in the United States of about .5, while a study of 
British imports 17 showed a price elasticity of approximately .64. 

If these elasticities are representative of price elasticities in general, it 
is apparent that the operation of the classical mechanism is even more 
difficult to explain than Taussig had supposed. Not only did the trade 
balances move with surprising rapidity, but they moved in the expected 
direction despite the fact that the physical volume of imports is normallv 
responsive only in a slight degree to changes in relative prices. In order to 
attribute the observed adjustments to changes in relative prices, it would 
in many instances be necessary to assume that demand elasticities are 
much higher than those which have actually been measured. 


Although the empirical evidence accumulated during the interwar 
period had clearly indicated the need for a reconsideration of the balance- 
of-payments mechanism, no substantial revisions of the accepted theory 
were made until Keynes published his General Theory. Thereafter, the 
missing link in the classical theory became almost self-evident: the rapid 
adjustment of a country's balance of payments which Taussig had ob- 
served, and which seemed to occur without the assistance of price changes 
or changes in central bank policy, was found to be largely the result of 
induced movements of income and employment. Suppose, for example, 
that Country A increases its imports from B, and that a deficit thus arises 
in A's balance of payments. The deficit may initially be financed by gold 
shipments or by a movement of short-term balances, but regardless of the 
method of financing, a more or less automatic mechanism will soon offset 
at least part of the initial disturbance. Income and employment will ex- 
pand in the export industries of B; the demand for home goods will there- 
fore rise in that country, and the expansion will spread from the export 
industries to the entire economy. As output and employment increase, 
Country B will increase its imports from A, thereby offsetting a part, or 
perhaps all, of the initial rise of exports to A. 

18 Randall Hinshaw, "American Prosperity and the British Balance-of-Payments Prob- 
lem," Review of Economic Statistics, February 1945, XXVII, p. 4. 

17 Tse-Chung Chang, "The British Demand for Imports in the Inter- War Period," 
Economic Journal, June 1946, LVI, p. 197. 


This, in brief outline, is the revised theory of the balance of payments 
which grew out of the theory of employment. Although the new theory, 
in complete form, was first presented by Mrs. Robinson and by Harrod in 
works previously cited, its main features can be found as early as 1936 in 
a remarkable article by Paish. 18 

Perhaps the most important single feature of the new concept is its 
comparative independence from banking policy. The cumulative move- 
ments of output and- employment which account for a large part of the 
adjustment of the balance of payments will normally be influenced only 
to a small extent by central bank action; to a much greater extent such 
income movements are a direct consequence of changes in the demand 
for goods and services. In the preceding illustration, for instance, if the 
initial surplus in Country B were offset by a gold inflow into that country, 
the central bank might attempt to neutralize or sterilize the gold. In other 
words, the banking authorities might prevent the gold inflow from in- 
creasing either the reserve ratios of the banks or the amount of money in 
circulation. This they could easily do by selling securities. Unless domes- 
tic investment were highly sensitive to a change in interest rates, how- 
ever, such action would not stop the rise of employment which was 
initiated in the export trades, and the adjusting process would accord- 
ingly proceed as before. The divorcing of the modern balancing mecha- 
nism from bank policy explains why a balancing tendency between 
foreign payments and receipts is sometimes apparent even when banks at 
home and abroad are carrying out neutralizing operations. Bank policy, 
apart from its influence on capital movements, can affect the balance of 
payments only through the circular flow of income, and the relation of 
bank policy to the circular flow is at best tenuous and uncertain. In the 
words of P. B. Whale: 

Since gold movements (or more generally, changes in reserves) and discount rate 
adjustments are displaced from their central position in the process of international 
price adjustment, the question of "observing the rules of the game", as this is ordi- 
narily understood, loses much of its importance. 19 

In short, a central bank which attempts to stabilize by offsetting rising 
exports with a sale of securities is not really interfering much with the 
"natural" balancing mechanism. But neither is it achieving much sta- 

18 F. W. Paish, "Banking Policy and the Balance of International Payments," Eco- 
nomica, November 1936, New Series III, pp. 404-422. 

19 P. B. Whale, "The Working of the Pre-War Gold Standard," Economica, February 
*937> New Series IV, p. 31; cf. also League of Nations, op. cit., Ch. IV. 



In the foregoing account of the balancing mechanism, the adjustment 
of international payments and receipts through changes in real income 
and employment has been referred to repeatedly as the "new" or "re- 
vised" or "modern" theory. Although this theory undoubtedly contains 
significant elements of innovation which justify calling it a new or mod- 
ern theory, it also has, like other scientific innovations, important ante- 
cedents. Indeed, after the publication of Viner's comprehensive studies 20 
it is now clear that even many of the English economists who are com- 
monly regarded as members of the classical school subscribed to a theory 
of adjustment which differed considerably from the classical theory, and 
had much in common with the modern view. Ricardo, for example, be- 
lieved that some disturbances to the international balance, such as an 
increase of agricultural imports resulting from a crop failure at home, 
could be rectified without gold movements and corresponding price 
changes. Although his reasoning on this point was somewhat obscure, 
Wheatley, as Viner shows, 21 gave an account of the same type of adjust- 
ment which indicated clearly how the international accounts might be 
balanced without gold movements and price changes. Wheatley argued 
that if England increased her agricultural imports because of a crop fail- 
ure, this in itself would increase the incomes of exporters to England, and 
that the ability of such exporters to purchase English goods would there- 
fore be greater than before, even without price changes. To some extent, 
in other words, the balance of payments tended to adjust itself by means 
of changes in purchasing power at home and abroad. A similar view was 
presented later (1840) by Longfield, 22 and still later (1889) by Basta- 
ble, 23 who applied this purchasing power theory to the disturbance re- 
sulting from the payment of a loan. Bastable argued that a payment from 
Country A to Country B would automatically increase the purchasing 
power of the receiving country and reduce the purchasing power of the 
paying country. Imports of A would therefore fall while imports of B 
would rise, even without price changes, and Bastable believed that the 
paying country could thereby achieve an export surplus equal to the 
annual payments, without gold movements. During the interwar period, 
ideas similar to Bastable's appeared in the well-known theory of Ohlin 24 

20 Jacob Viner, Studies in the Theory of International Trade (New York, 1937), 
especially Ch. VI and VII. 

21 Void., pp. 295-297. 

22 Void., p. 297. 
M lbid., pp. 302-303. 

24 Bertil Ohlin, "Transfer Difficulties, Real and Imagined," Economic Journal, June 
1929, XXXIX, pp. 172-178. 


and other Scandinavian economists that reparations and similar transfers 
can be carried out by means of shifts in purchasing power, and that no 
price movements need occur. 

All of these purchasing power arguments sound surprisingly like the 
theory discussed earlier in the present review. What, then, is the justifi- 
cation for calling the adjustment through changes in income a new or 
modern theory? Wherein does it differ from the theories of Wheatley, 
Ricardo, Bastable, Ohlin, and others? The difference, in my opinion, is 
primarily that the earlier expositions lacked a theory of employment or 
income, and were therefore unable to explain just how far the adjusting 
process could go. Some of the earlier explanations were vague and am- 
biguous as to the extent of income movements, while the later ones were 
frequently erroneous. There was a strong tendency in the later discus- 
sions, for example, to cling to the assumption that full employment pre- 
vails at all times, and to assume, therefore, that in the case of a money 
transfer, purchasing power is increased in the receiving country and re- 
duced in the paying by exactly the amount of the transfer. In the words 
of Iversen, ". . . the total amount of buying power in the two countries 
together is unchanged; only its distribution between them is changed." 25 
In the light of the modern theory of employment, it is obvious that this 
doctrine of the conservation of purchasing power, which was an integral 
part of a number of the pre-Keynesian discussions, cannot be supported. 
When secondary as well as primary changes in income have been taken 
into account, it is clear that something more than a mere shift in pur- 
chasing power has occurred; in addition, there may be a net change in 
output and employment both at home and abroad. It is the ability to set 
limits to these changes in purchasing power, or at any rate to determine 
the conditions on which the changes depend, which distinguishes the 
new theory from the older shifts-of-purchasing-power doctrine. From this 
point of view, the theory of Bastable and Ohlin is a stepping stone to the 
new theory, but is not in itself a complete explanation of how balances of 
payments are affected by changes in income. 


If the modern theory establishes more definite limits than its predeces- 
sors to the balancing influence of income movements, what are these 
limits? In particular, is the theory of employment a complete explanation 
of the balancing process or is it only a partial explanation? In order to 
give complete answers to these questions, it would be necessary to con- 

25 Carl Iversen, Aspects of the Theory of International Capital Movements (London, 
1936), p. 232. 


sider the components of national income in considerable detail. The 
following remarks will therefore be limited to a summary of general con- 
clusions and to a statement of the opinions of a number of economists. 

The balancing process is closely related to what might be called "the 
fundamental income identity for an open economy." This identity states 
simply that, for any individual country, savings are the sum of two com- 
ponents: (1) net domestic investment; (2) the balance of payments on 
current account. Although this is an identity, being simply a definition of 
savings over any past accounting period, it may also be regarded as a condi- 
tion of equilibrium, provided all the components are interpreted as in- 
tended savings, intended investment, etc. Let us see, now, how the income 
identity can be applied to our earlier discussion of the balance of payments 
between Countries A and B, when this balance is disturbed by an in- 
creased demand in A for the products of B. Consider the situation in B, 
the country which initially has an increase in exports. Since income rises 
in B, we may take it for granted that savings, interpreted in the intended 
sense, will also rise. From the savings-investment relation it follows that 
the sum of domestic investment plus the balance on current account must 
also be higher than before. Thus, net domestic investment must be 
higher than in the initial position, or the balance on current account 
must be more favorable to B, or some combination of these two must 
occur. Which outcome is most likely? In the earlier discussions of the 
balancing process by Mrs. Robinson and R. F. Harrod, there was a tend- 
ency to take the level of investment as given, and to consider only the 
influence of saving and consumption on the balance of payments. Under 
these assumptions, the balancing process is obviously incomplete. Unless 
domestic investment increases in B, for example, savings can remain 
above the previous level only to the extent that the balance on current 
account remains more favorable to that country. The induced rise of in- 
come in B will thus offset a part, but not all, of that country's surplus on 
current account. 

Later discussions modified this view somewhat by showing that invest- 
ment, in the short run at any rate, may depend upon the level of income, 
and that induced cumulative movements of income may accordingly be 
large enough to offset a balance-of-payments disturbance completely. 26 
While some differences of opinion still exist concerning the role of in- 
duced investment, the conclusion of most economists seems to be that, 

28 See, e.g., Lloyd A. Metzler, "Underemployment Equilibrium in International Trade," 
loc. cit., 'passim; League of Nations, op. tit., Ch. IV, Sec. 5; Ragnar Nurkse, "Domestic 
and International Equilibrium," in Chapter XXI of The New Economics, Seymour E. 
Harris, ed. (New York, 1947). 


except under unusual conditions, the adjustment of a country's balance 
of payments by means of income movements is likely to be incomplete. 


Like the classical theory of the balance of payments, the theory which 
has emerged in the last ten years envisages a more or less automatic bal- 
ancing mechanism. Unlike the classical theory, however, the new expla- 
nation, as we have seen, normally accounts for only a part of the adjust- 
ment and thus constitutes a theory of disequilibrium as well as a theory 
of equilibrium. Moreover, the cumulative movement of income at home 
and abroad which is the essence of the modern theory will not occur 
unless the disturbing influence affects the circular flow of income as well 
as the balance of payments. The adjustment of a country's balance of 
payments to speculative capital transfers or other disturbances which 
have no direct effect upon the circular flow of income is thus likely to be 
slow and insignificant. On the other hand, if the initial disturbance is an 
increase or decrease of direct investments abroad or any other event 
which alters the flow of income, the secondary adaptation of the balance 
of payments to the new conditions will probably be substantial. In this 
respect, as in others, the new theory differs from the classical, for the 
adjustment envisaged by the classical theory was much the same regard- 
less of the nature of the initial disturbance. 

Perhaps the most important difference, however, is in the nature of the 
adjustment itself. In the modern view, a country with a deficit in its 
balance of payments is likely to eliminate this deficit, in part at least, 
through a low level of income and employment. The conflict between 
domestic stability and international equilibrium, which has long been a 
familiar part of classical monetary theory, is thus shown to be much more 
important than had formerly been supposed.' In an unstable world, the 
choice confronting an individual country is not merely between price 
stability and international equilibrium, as envisaged by the classical 
theory, but between stability of employment and international equilib- 
riums In recent years there has been a growing recognition of this conflict 
and the difficulties of resolving it. The necessity for international co- 
operation to ensure a balanced and stable rate of economic growth 
throughout the world has thus become increasingly apparent. 

It cannot be said that much has yet been accomplished in reaching this 
objective. Nevertheless, a tremendous change is evident almost every- 
where in the attitude of individual countries toward control of the rate of 
economic activity. If international planning for stability is not yet popu- 
lar, many countries at least are making plans of their own for stabilizing 


their economies and using their resources fully. It remains to be seen how 
successful the new programs will be, once the prolonged period of transi- 
tion from war to peace has been passed. It is my own judgment, however, 
that most of them, while not perhaps eliminating all fluctuations of eco- 
nomic activity, will probably eliminate the large movements of employ- 
ment which we associate with major business cycles. If so, the conditions 
under which international trade is carried on in the future will be entirely 
different from the conditions of the interwar period, and probably some- 
what different also from the conditions before the First World War. 
Fluctuations of demand arising from movements of income will be rela- 
tively small, and resources will be largely employed, as postulated by the 
classical theory of the balance of payments. This means, among other 
things, that induced movements of output and employment, such as those 
which have explained a part of the balancing of international accounts in 
the past, will probably not be permitted in the future. We have thus 
reached the somewhat paradoxical result that the more successful 
Keynesian remedies prove to be in solving problems of domestic stability 
the less need we shall have for Keynesian economics in describing inter- 
national affairs. 27 

What, then, will be the mechanism of adjustment in the future? If 
induced changes in employment are prevented or greatly reduced, vir- 
tually the only method of balancing international accounts without resort 
to direct controls will be through changes in the terms of trade, i.e., 
through the price system. This does not mean, however, that the classical 
mechanism of price adjustments will experience a renaissance, for coun- 
tries which adopt policies to stabilize output will no doubt be equally 
interested in stabilizing the general level of prices and costs. It is there- 
fore not to be expected that general price movements wall supplant move- 
ments of output as regulators of the balance of payments. Although the 
classical theory, in the strict sense, will thus be as outmoded as the mod- 
ern theory, the method of adjustment which finally evolves will probably 
be more nearly akin to the classical mechanism than to the modern. Even 
without general price and cost changes, the essential means of adjust- 
ment contemplated by the classical theory— a change in the terms of trade 
—can be accomplished through changes in exchange rates, and if the 
present trend toward widespread state control of trade is to be halted the 
international monetary system will have to move increasingly toward 
such an arrangement. Indeed, in a world of high and stable employ- 

27 It is perhaps unfair to describe the modern theory of adjustment of the balance of 
payments as "Keynesian economics/' since Keynes himself had little to do with it. The 
new theory is Keynesian only in the sense that it is a direct outgrowth of the theory of 


ment, movements of exchange rates are virtually the only more or less 
automatic means of influencing international trade without resorting to 
direct controls. For this reason, it seems appropriate to review, in the sec- 
tion which follows, the developments during the interwar period in the 
theory of fluctuating exchange rates. 

III. Fluctuating Exchange Rates 


During the First World War, the gold standard was suspended 
throughout the world, and although most countries eventually returned 
either to the gold standard or to the gold exchange standard after the war 
had ended, the resumption of gold payments was a long and protracted 
process. Throughout most of the decade of the twenties there were ac- 
cordingly substantial fluctuations in the external values of many curren- 
cies. Moreover, the process of stabilization had hardly been completed 
when a large part of the world once more abandoned the gold standard 
as a consequence of the Great Depression in the early 'thirties. Later, 
during one period in the middle thirties, currency values were relatively 
stable without a formal return to the gold standard, but this stability was 
again disrupted, this time by the United States depression of 1937-38 
and by the abnormal capital movements which preceded the outbreak of 
the Second World War. The interwar period was thus a period of fluctu- 
ating exchange rates; only a few years during the entire period were 
characterized by exchange-rate stability. 28 

It was natural, under such conditions, that economists should have de- 
voted considerable attention to the effects of exchange fluctuations and 
that permanent contributions should have been made, in consequence, 
to this particular branch of the theory of international trade. The theoreti- 
cal development, however, did not proceed at a uniform rate throughout 
the interwar years. The underlying causes of exchange fluctuations in the 
'twenties were quite different from the causes of the later exchange move- 
ments, and the theories developed in the two periods were likewise dif- 
ferent. Movements of exchange rates in the early 'twenties were largely 
an aftermath of the war. Postwar inflation had brought about marked dis- 
parities in the internal price levels of different countries, and the ex- 
change-rate movements of this period were principally a reflection of 

28 One of the best descriptions of exchange-rate movements during the interwar period 
as a whole is in the League of Nations study previously cited. (International Currency 
Experience: Lessons of the Interwar Period, Ch. V). For an excellent account of the 
exchange situation in the early 'thirties, see Seymour E. Harris, of. cit., passim. 


these price movements. Indeed, inflation and the resulting price differ- 
ences played such a dominant part in the determination of exchange rates 
during this period that a distorted theory of exchange rates enjoyed wide 
popularity. This was the theory of purchasing power parity, which at- 
tributed changes in exchange rates entirely to relative movements of 
internal purchasing power. 29 Even at the height of its popularity, how- 
ever, the parity theory was a target of severe criticism, and eventually it 
was almost completely discredited as an explanation of exchange rates. 30 
There is no need, in a review such as this, to discuss the criticisms in 
detail. The inability of the parity theory to allow for shifts in interna- 
tional demand, for capital movements, for technological changes, or for 
any other events altering the terms of trade soon made it apparent that 
the theory was not a general explanation of exchange rates, but was ap- 
plicable only under special conditions. 31 There were also other criticisms 
of a more technical nature, such as the difficulty of selecting an appro- 
priate index of prices or costs, but these need not concern us here. 

The movement away from the parity theory was accelerated in the 
early 'thirties, when the Great Depression forced most countries off the 
gold standard once more, and when exchange rates were subjected to in- 
fluences which clearly could not be explained by price movements 
alone. 32 The balance-of-payments difficulties of the depression years were 
principally attributable to the fact that the depression did not affect the 
demand for all countries' exports uniformly. Although induced move- 
ments of real income tended, to some extent, to redress the balance, as 
described in the preceding section, these income movements did not 
effect a complete adjustment. 

From the point of view of economic analysis, one of the most impor- 
tant results of the experience with fluctuating exchange rates during the 
'thirties was a profound skepticism concerning the effectiveness of ex- 
change-rate adjustments in rectifying a balance-of-payments discrepancy. 
This skepticism was partly a consequence of certain special conditions of 
the 'thirties which are not likely to be repeated in the future, but it was 

29 See Gustav Cassel, Money and Foreign Exchange after 1914 (New York, 1922), 
pp. 137-186. 

30 One of the definitive accounts is that of C. Bresciani-Turroni, "The Purchasing 
Power Parity Doctrine," L'Egy-pte Contenrporaine, 1934, pp. 433-464. 

31 Cf. Jacob Viner, Studies in the Theory of International Trade, pp. 379-387. In my 
opinion, the criticism of the parity doctrine went too far, and the theory was rejected 
even for situations in which it was valid. During the 'twenties, for example, disparities 
in price movements between countries were clearly the most important influences on 
exchange rates, and purchasing power parity was therefore a useful doctrine. See Lloyd 
A. Metzler, "Exchange Rates and the International Monetary Fund," in Postwar Studies 
No. 8, Board of Governors of the Federal Reserve System (Washington, 1947). 

32 See Seymour E. Harris, op. cit., passim, but especially Ch. IV. 


also a consequence, as we shall see, of more fundamental difficulties with 
the balance-of-payments mechanism. Consider first the special conditions 
of the 'thirties. 

The adjustment of exchange rates in the 'thirties was complicated both 
by large-scale speculative capital movements, which added to the insta- 
bility of exchange rates, and by competitive devaluation, which reduced 
the effectiveness of depreciation for the deficit countries. Although these 
complications created serious doubts regarding the benefits of flexible ex- 
change rates, neither of them presents an insurmountable obstacle to a 
flexible exchange system. Moreover, there are good reasons for supposing 
that such disturbing events, in the future, will be entirely prevented or 
at any rate greatly reduced. Under the Articles of Agreement of the In- 
ternational Monetary Fund the member countries have committed them- 
selves, in effect, to submit the question of exchange-rate adjustments to 
international collaboration. Changes in the par value of a currency are 
to be made only when necessary to correct a fundamental disequilib- 
rium, 33 and this presumably means that a member of the Fund will not 
be able to devalue its currency unless it has a persistent deficit in its 
balance of payments. 34 

If devaluation is limited to deficit countries, the degree of such de- 
valuation will likewise be limited largely to the amounts needed to restore 
equilibrium, and will not be affected, to the extent that it has been in 
the past, by speculative capital transfers. During the war, and even 
before, comprehensive exchange controls were adopted throughout the 
world, and up to the present time most of these controls have been re- 
tained. While controls of foreign exchange received on current account 
will eventually be removed, under the Fund agreement, there is no 
commitment to remove controls of capital movements, and it is generally 
believed that such capital controls will continue in force. Indeed, under 
certain conditions the Fund itself may require a member country to 
adopt controls of capital exports. 35 It will be impossible, of course, to 
control or prevent every undesirable capital transfer, since some transfers 
can be disguised as export transactions. Nevertheless, the bootleg trans- 

33 Articles of Agreement of the International Monetary Fund, Article IV. 

34 See Ragnar Nurkse, Conditions of International Monetary Equilibrium, Princeton 
University, Essays in International Finance, No. 4 (Princeton, 1945); see also Gottfried 
Haberler, "Currency Depreciation and the International Monetary Fund," Review of 
Economic Statistics, November 1944, XXVI, pp. 178-181; and Alvin H. Hansen, "A 
Brief Note on 'Fundamental Disequilibrium,'" ibid., November 1944* XXVI, pp. 182- 

35 Article VI, Section 1 (a) of the Article of Agreement provides that "a member may 
not make net use of the Fund's resources to meet a large or sustained outflow of capital, 
and the Fund may request a member to exercise controls to prevent such use of the 
resources of the Fund." 


fers which manage to evade control will clearly be much smaller and less 
disruptive than were the speculative capital transactions of the 'thirties. 


Although some of the most troublesome features of the fluctuating 
exchange rates during the 'thirties can thus probably be prevented in the 
future, one fundamental problem remains to be discussed before con- 
cluding that exchange-rate adjustments are an effective way of balancing 
international payments and receipts. Even in the absence of speculative 
transactions there is some doubt as to whether currency depreciation, in 
the short run at least, can eliminate or reduce a deficit in a country's 
balance of payments. Exchange-rate movements affect the principal items 
in a country's balance of payments— exports and imports— primarily by 
altering the ratio of domestic to foreign prices, and if elasticities of de- 
mand are small, such relative price movements may be ineffective or 
may even affect the balance of payments adversely. In other words, the 
questions which were asked in an earlier section concerning the operation 
of the classical gold-standard mechanism under conditions of inelastic 
demand must be asked again with regard to fluctuating exchange rates. 

In classical and neo-classical economics the possibility of exchange 
fluctuations having a perverse effect was seldom if ever discussed, but 
during the interwar period the question assumed increasing importance. 
Empirical investigations, on the one hand, revealed that price elasticities 
of demand were much smaller than had usually been assumed; and on 
the other hand, the experience of certain countries with depreciation, 
particularly producers of primary products, led these countries to doubt 
the effectiveness of this method of increasing the value of exports. 
The possibility that flexible exchange rates might be inherently un- 
stable, a fall in the price of a currency increasing rather than reducing 
that country's deficit, was thus widely discussed, and an important con- 
tribution to the theory of exchange stability was gradually developed. 
Before considering the part played by individual economists in this 
development, however, a brief account will be given of the present status 
of the theory itself. 

Stability of market exchange rates, like the stability of any price 
system, requires that a fall in the price of a particular currency shall 
reduce the excess supply of that country's currency on the foreign 
exchange markets, or that a rise in price shall reduce excess demand. A 
theory of exchange stability, based upon this principle, was developed 
during the interwar years, but the theory remains in a relatively ele- 
mentary state. It is customary, and indeed necessary in the incomplete 


state of our knowledge, to discuss the effects of depreciation for the 
simplified case of two countries trading in only two commodities. Even 
with this simplification the theory of exchange stability, as it was worked 
out during the interwar period, remains somewhat complicated. Since 
most of the basic conclusions are simple, however, it seems better to 
state these conclusions in the form of a few categorical remarks than 
to present the cumbersome algebra. In what follows, the terms "exports" 
and "imports" will be understood to include all of the items in a country's 
foreign receipts and payments on current account. 

If the demand for both exports and imports is inelastic, depreciation 
normally reduces a country's foreign-exchange receipts as well as its 
disbursements. The physical volume of exports is increased, of course, 
but the increase in volume does not compensate for the decline in 
foreign price, and the foreign-exchange value of exports accordingly 
declines. With respect to imports, both the physical volume and the 
foreign price decline to some extent, and depreciation thus reduces 
expenditures of foreign currency no matter how small the elasticity of 
import demand may be. The final effect upon a country's balance of 
payments therefore depends upon the magnitude of the decline in the 
foreign value of exports compared with the decline in the value of 
imports. The balance of payments will not be improved unless the 
value of imports falls more than the value of exports. While it is con- 
ceivable that this may occur even when the demand for both exports 
and imports is inelastic, it is not likely to occur if such elasticities are 
exceedingly small. 

Most economists who have considered the problem of exchange 
stability have presented what might be called an "elasticity of the balance 
of payments." Consider only two countries, Y x and Y 2 , and let ^ and rj 2 
be the elasticities of demand for imports in the two countries. Similarly, 
let e x and e 2 be the elasticities of supply of exports. If the discrepancy 
between exports and imports is small, relative to the total value of foreign 
trade, it can easily be shown that a devaluation of the currency of either 
country in the proportion K will bring about a change, positive or 
negative, in that country's balance of payments on current account, 
which has the following value, relative to the value of exports: 

vi ™ 2 + e i + e 2 ) + e^ Oh + y 2 — ) 
| Oh + e 2 ) (% + ej ) 

The foreign exchange market is obviously unstable unless the expression 
in brackets is positive, for exchange stability requires that depreciation 
must increase a country's net supply of foreign exchange. If the supply 


schedules of exports are positively sloped while the demand schedules 
for imports are negative, all of the elasticities of supply and demand in 
the above expression will be positive. 30 From this it follows that the 
elasticity of the trade balance cannot be negative unless ^ + rj 2 . — - I is neg- 
ative and large. A sufficient condition for stability is thus that the sum 
of the two demand elasticities shall be greater than unity. Even if this 
sum is smaller than unity, the elasticity of the trade balance may still be 
positive if the supply elasticities, e x and e 2 , are sufficiently small. 

Since stability depends upon supply elasticities as well as demand 
elasticities, it may be useful to consider two limiting cases. First, if ex- 
ports are produced under constant supply prices, as they are for many 
manufactured products, both e x and e 2 are infinite, and the elasticity of 
the balance of payment becomes rj x + V2 — i« The minimum require- 
ment for stability in this case is thus that the sum of the two demand 
elasticities shall exceed unity. At the other extreme, where the supply 
of exports is completely inelastic, as it is in the short run for certain 
agricultural products, the elasticity of the balance of payments is always 
positive and has a value of unity, regardless of the demand elasticities. 
Under such conditions, depreciation always improves a country's bal- 
ance of payments no matter how inelastic the demands for imports 
may be. 

The foregoing conclusions are the principal technical results of the 
interwar discussion of exchange stability. When we attempt to apply 
these results to actual problems and to form a judgment as to the effects 
of exchange-rate movements, we are confronted, unfortunately, with 
a serious lack of empirical evidence. Almost no information is available 
concerning supply elasticities, and estimates of demand elasticities are 
available only for a few countries. Nevertheless, such information as 
we do have indicates rather small demand elasticities— estimates of .5 for 
the United States and .64 for the United Kingdom have already been 
mentioned. If such elasticities are typical, we are forced to conclude, 
I believe, that fluctuations of exchange rates are not likely, in the short 
run at least, to bring about an appreciable improvement in a country's 
balance of payments. But perhaps this is too pessimistic. It is no doubt 
true that the demand for imports is frequently inelastic, but in the 
short run the same is true of the supply of exports, particularly of 
agricultural exports, and the theory of exchange stability discussed 
above has shown that inelastic export supplies may conceivably com- 

88 Demand elasticities are here defined, in the Marshallian manner, as t~ • 

J dp x 

whereas supply elasticities are defined as -5— • -, without the change in sign. 


pensate for the unfavorable effects of inelastic demand. Even if the 
reaction of the balance of payments is favorable to the depreciating 
country, however, it is likely to be small in magnitude. If the demand 
for imports is highly inelastic, as it normally is in the short run, the 
elasticity of the balance of payments will likewise be inelastic, unless 
the supply of exports is completely unresponsive to changes in price. A 
small change in the balance of payments, relative to the value of exports, 
may therefore require a very large proportionate change in exchange 

If we consider the problem over a longer period of time, the prospects 
for exchange depreciation are more promising. In the short run the 
demand for imports is inelastic primarily because both producers and 
consumers cannot rapidly adapt their purchasing habits or methods of 
production to a change in relative prices. In the long run, however, the 
possibilities of substitution between domestic and foreign products or 
raw materials is considerably greater. It seems probable, therefore, that 
the long-run elasticity of demand for imports in most countries is larger 
than the statistical evidence, based upon short-run conditions, would 
indicate. If a country with a balance-of-payments deficit depreciates its 
currency, a considerable improvement in its balance of payments may 
eventually occur, even though the immediate effects are small. During 
the interwar years, the long-run influences were for the most part nulli- 
fied by competitive devaluation; hardly any of the deficit countries had 
an opportunity to test the effects of cheaper currency over a protracted 
period of time. The interwar experience therefore gave a somewhat 
distorted view of the effects of depreciation. 


The foregoing account of exchange fluctuations and the balance of 
payments has been presented without any reference to the economists 
who were responsible for this line of thought. Actually, the conditions 
of exchange stability discussed above were discovered independently 
during the interwar period by three different economists. So far as I am 
aware, the correct conditions of exchange stability appeared first in the 
Economic Journal of 1920, in a brief and unfortunately neglected note 
by C. F. Bickerdike. 37 At the time the note was published, the pound 
sterling was an inconvertible currency. Currency controls which had 
been introduced during the war were removed in the early part of 191 9, 
and as a result the dollar price of the pound sterling declined sharply, 

37 C. F. Bickerdike, "The Instability of Foreign Exchange," Economic Journal, March 
1920, XXX, pp. 1 18-122. 


from $4.76 in February of that year to $3.81 in December. A large part 
of this depreciation was attributable to the abnormal British demand for 
imports at the end of the war, and to the discrepancy which had devel- 
oped during the war between the British and American price levels. 38 
Nevertheless, the violence of the decline in the price of sterling led to 
a considerable amount of discussion in the United Kingdom as to whether 
free exchange markets for inconvertible currencies were not inherently 
unstable. In other words, the British began to doubt that currency de- 
preciation would reduce the deficit in their balance of payments, and 
suspected, on the contrary, that such depreciation might only make the 
situation worse. Bickerdike's note was, to my knowledge, the first scien- 
tific expression of these doubts. Bickerdike derived a formula for the 
relation between depreciation and the balance of payments which was 
essentially the same as the one given above, and he then gave a pessi- 
mistic interpretation to his results: 

With the prospect of inconvertible paper money in many countries for a consider- 
able time, it is important to recognize that a high degree of instability of exchange 
rates is almost inevitable, and is not solely due to the continual increase of such 
money to which Governments have been obliged to resort. The question may be 
looked at from the point of view of very short periods, such as day to day, or rather 
short periods, such as a year, or over considerable periods of years. In each case a 
consideration of the circumstances leads to the conclusion that a high degree of 
instability is to be expected with inconvertible paper currencies. 39 

Bickerdike pointed out, as we have shown above, that in the short run 
the foreign and domestic elasticities of demand are likely to be small, and 
that modest balance-of-payments deficits may therefore produce violent 
changes in exchange rates. It is perhaps arguable, in this regard, that 
he did not attach sufficient importance to the stabilizing influence of 
inelastic short-run export supply. In any event, his work was largely 
forgotten after the postwar period of exchange fluctuations had ended, 
and the subject was not revived again until the 'thirties. 

The second presentation of the theory of exchange stability was in 
Mrs. Robinson's well-known essay on the foreign exchanges, published 
in 1937 at the end of another period of fluctuating exchange rates. 40 
Although Mrs. Robinson's method of presentation was somewhat differ- 
ent, her analytical results were the same as Bickerdike's, and she reached 

38 In 1919, the British wholesale price index stood at 242 (1913 = 100), compared 
with an index of 206 for the United States. (League of Nations, Statistical Yearbook.) 

30 C. F. Bickerdike, o?. ciu, p. 118. 

40 Joan Robinson, "The Foreign Exchanges," in her Essays in the Theory of Employ- 
ment, pp. 188-201. 


similar conclusions. She placed more emphasis than had Bickerdike, 
however, on the stabilizing effects of low supply elasticities. She pointed 
out, for example, that the inelasticity of the Australian wool supply was 
an important factor in the benefit which that country derived from 
depreciation in 1931. 

The third treatment of the subject of exchange stability was by A. J. 
Brown. 41 Except for a complication introduced by the presence of im- 
ported raw materials in exports, 42 Brown's analytical results were the 
same as those of Bickerdike and Mrs. Robinson. His interpretation of 
the results, however, was more optimistic. He argued, in particular, that 
the foreign elasticity of demand for British exports is likely to be high, 
since Britain is competing with other countries in these foreign markets, 
and depreciation would enable her to take customers away from her 
rivals. This is a question of considerable importance for the theory of 
exchange rates, and it is a question which, in my judgment, has not been 
satisfactorily answered. It is frequently said, by analogy with the theory 
of perfect competition, that a country which is small, or a country whose 
exports constitute a small part of the world market, can improve its 
balance of payments by means of depreciation, since the demand for 
this single country's products will normally be highly elastic. 43 From 
this it is then sometimes concluded that discussions of exchange stability 
which consider only two countries involve a pessimistic bias, and that if 
a world economy consisting of many countries were taken into account 
the probability of exchange stability would be appreciably greater. 

Graham raised the same point in a slightly different way in two articles 
in the Quarterly Journal of Economics. 44 Although he did not state 
whether he was considering a system of flexible exchange rates or the 
classical gold-standard mechanism, his argument is applicable to either 
case, since he dealt with the barter or reciprocal-demand diagrams of 
neo-classical economics. He argued that the classical analysis in terms 
of two commodities and two countries exaggerated the instability of 
the terms of trade. With a wide variety of exports, or with several 
countries participating in trade, Graham believed that the terms of 
trade, or the exchange rates, would be confined within rather narrow 

41 A. J. Brown, "Trade Balances and Exchange Stability/' Oxford Economic Papers, 
April 1942, No. 6, pp. 57-76. 

42 His treatment of raw materials contained an error. He argued that a high proportion 
of foreign raw materials in exports is a stabilizing factor. Although space does not permit 
a demonstration here, it can be shown that this is incorrect, and that the contrary is true. 

43 See, e.g., Seymour E. Harris, Of. cit., pp. 62-66. 

44 Frank D. Graham, "The Theory of International Values Re-examined," Quarterly 
Journal of Economics, November 1923, XXXVIII, pp. 54-86; idem, "The Theory of 
International Values," ibid., August 1932, XLVI, pp. 581-616. 


limits, through the substitution of the products of one country for those 
of another. In his own words, ". . . any alteration in the rate of inter- 
change will affect the margin of comparative advantage of some country 
in the production of some one of the commodities concerned, will bring 
that country in as an exporter where formerly it was an importer, or 
as an importer where formerly it was an exporter, according as the terms 
of trade move one way or the other . . ," 45 

While these substitution effects on the supply side undoubtedly exert 
a stabilizing influence, it seems to me that Graham has overstated his 
case. The mere existence of a large number of trading countries or a 
large number of commodities will not stabilize the exchange markets 
unless there is a wide variation in cost conditions among the different 
countries. In other words, competition, by itself, is no guarantee of 
stability. Suppose, for example, that three countries, B, C, and D, were 
in close competition in the import market of A. The world demand for 
the products of any one of these countries would then be highly elastic; 
if B depreciated, for instance, she could thereby increase her exports to 
A at the expense of C and D. But this would involve a deterioration in 
the balances of payments of C and D which, in turn, would normally 
lead to exchange adjustments in these latter countries, and when all 
such secondary adjustments have been taken into account it is by no 
means clear that the initial depreciation by B would have reduced the 
deficit in that country's balance of payments. In order that substitution 
between products shall stabilize the exchange markets, it is not only 
necessary that B, C, and D shall be in close competition in A, but also 
that A shall be in close competition with at least one of these countries 
in the markets of a third country. 

Graham was well aware of this fact, but he believed that the type of 
"linked" competition which he postulated in his numerical examples 
was fairly common in the actual world. In other words, he felt that 
a number of countries would always change from exporters to importers 
of particular commodities with slight movements in the terms of trade. 
With the world divided as it has been in the past between exporters of 
raw materials and exporters of manufactures, Graham's supposition of 
continuous variation seems to me too optimistic. Indeed, our experience 
during the Great Depression has shown that enormous deterioration can 
occur in the terms of trade of an agricultural country without materially 
altering the character of that country's exports or imports. The substitu- 
tion effects which Graham discusses are thus slow to occur and frequently 
do not take place rapidly enough to offset the adverse effects of low 

45 Idem, "The Theory of International Values Re-examined," loc. cit., p. 86. 


elasticities of demand. 46 In the short run therefore, and perhaps even over 
periods of time as long as five to ten years, low elasticities of demand may 
be a serious source of instability in foreign-exchange markets, even with 
a large number of countries competing in world markets. 


Both the interwar experience with fluctuating exchange rates and the 
theory of exchange stability which emerged during the interwar years 
have clearly shown that adjustments of exchange rates are not likely, 
in the short run, to be an efficient or effective means of eliminating a 
deficit or a surplus from a country's balance of payments. If the demand 
for imports is inelastic, as it appears to be in many countries, currency 
depreciation may have an insignificant or even a perverse effect upon 
a country's balance of payments. Considering the low price elasticities 
which have been found in most empirical studies of demand, it seems 
probable that depreciation, in the short run, cannot improve a country's 
trade balance unless the inelastic demand for imports is matched by a 
correspondingly inelastic supply of exports. Even in this case the elasticity 
of the trade balance will probably be small, and a substantial movement 
of exchange rates may therefore be required to eliminate rather modest 
deficits. In other words, over comparatively short periods of time, move- 
ments of exchange rates are not an efficient means of allocating resources 
between foreign and domestic use. 

This fact became increasingly apparent in both theory and practice 
during the interwar years, and it has now been explicitly recognized 
in the plans and institutions which have been developed for international 
trade in the future. Under the Articles of Agreement of the International 
Monetary Fund, exchange rates are stabilized, and exchange adjustments 
are to be made only occasionally in response to a fundamental dis- 
equilibrium in a country's balance of payments. In the short run, deficits 
are to be met by the use of foreign balances or, under the proposed 
International Trade Organization, 47 by the direct control of imports. 
While this procedure is clearly more realistic than some of the earlier 
proposals for a flexible exchange system, there is a danger that in 
recognizing the limitations of exchange adjustments we shall overlook 
their benefits. In the long run, when time has been allowed for the sub- 

46 In his second article, Graham recognized this possibility and conceded that the 
classical reciprocal demand equations, for the short run, might have considerable validity. 

47 Havana Charter for an International Trade Organization, United Nations Con- 
ference on Trade and Employment, held at Havana, Cuba, November 21, 1947 to 
March 24, 1948. Final Act and Related Documents (Havana, 1948), Articles 13, 14, 
15, and 23. 


stitution of one method of production for another, and when consumers 
have had an opportunity to adjust their spending to a change in relative 
prices at home and abroad, demand elasticities for imports will obviously 
be considerably greater than the elasticities computed for the interwar 
period. Movements of exchange rates are therefore likely to have a 
significant ultimate effect upon a country's balance of payments even 
though the immediate effects are small. The problem which confronts 
us today is how to preserve the long-run position of exchange adjust- 
ments while frankly recognizing the short-run limitations. To solve this 
problem, we must prevent the direct controls, which may be necessary 
in the short run, from becoming frozen into a permanent system of trade 

IV. Price Theory and International Trade 

Among economists, the interwar period will no doubt be remembered 
most vividly as the period of the Great Depression and the closely re- 
lated revolution in the theory of employment. Even without the 
Keynesian revolution, however, the period would have been one of ex- 
traordinary growth in economics, particularly in the field of general price 
theory; several important discoveries were made, during the interwar 
years, in such diverse subjects as the theory of consumer's choice, the 
theory of production, and the theory of monopoly and competition. Since 
price theory, or the theory of value, has always been intimately associated 
with international economics, it was not surprising that many of the 
innovations in price theory were eventually applied to the special prob- 
lems of international trade. The classical theory of the international 
price system was thus somewhat modified and modernized. Despite 
these modifications, however, many of the conclusions which the Eng- 
lish economists had reached with more antiquated equipment remained 
essentially unchanged. Because of space limitations, it is impossible to 
consider all of the recent changes in the theory of international prices. 
The following discussion is therefore limited to the innovations in three 
broad fields, the theory of demand, the theory of production, and the 
theory of general equilibrium. 


The theory of demand, or the theory of consumer's choice, enters into 
international economics primarily in the discussion of the gains from 
international trade. In the classical theory, the benefit which a countrv 
derived from specialization and trade was measured by the difference 


between the international rate of exchange of commodities and the 
rate which would have prevailed in the absence of international trade. 48 
In other words, the gain from trade was an objective quantity, indicating 
the saving in resources from specializing and trading rather than pro- 
ducing all commodities at home. One of Mill's great achievements, of 
course, was to show how the gain from trade, thus measured, is deter- 
mined by conditions of demand both at home and abroad. 49 Although 
Mill did not go beyond this conception, and did not attempt to relate 
his demand schedules to underlying utilities, he was nevertheless able to 
reach conclusions with respect to commercial policy which require very 
little modification even today. It remained for Marshall to apply the 
utility concept to international demand, and to measure the gain from 
trade by means of his well-known theory of producer's and consumer's 
surpluses. 50 This refinement, however, was not revolutionary, for Mar- 
shall's conclusions tended, in the main, to confirm those of Mill. Sub- 
sequent developments with regard to demand theory consisted principally 
of discarding the concept of measurable utility and substituting ratios 
of marginal utilities. With this innovation, attempts to measure the total 
gains from trade by means of Marshall's consumer's and producer's sur- 
pluses were discarded, but once more the fundamental principles re- 
mained largely intact. Prior to the interwar period, the principal modi- 
fications of the classical concept of 'gains from trade" thus consisted 
primarily in changes in the theory of demand. On the side of produc- 
tion, the labor theory of value, although largely discredited in general 
price theory, continued to be employed in the theory of international 

trade. 51 

Although Bastable had earlier altered the classical theory of production 
to some extent by introducing diminishing returns into his theory of inter- 
national trade, 52 perhaps the principal innovation in this regard was made 
by Haberler, who employed a production substitution curve to indicate 
the possible combinations of quantities of two goods which could be pro- 
duced with given quantities of the factors of production. This curve was 
defined in such a way that its slope at any given point represented the 

48 See David Ricardo, The Principles of Political Economy and Taxation, Gonner, ed. 
(London, 1891), Ch. VII. 

49 John Stuart Mill, Essays on Some Unsettled Questions of Political Economy, 3rd 
ed. (London, 1877), pp. 1-21. 

60 Alfred Marshall, Money, Credit and Commerce (London, 1923), Appendix J. 

51 See, e.g., F. Y. Edgeworth, Papers Relating to Political Economy, Vol. II (London, 
1925), pp. 42-45; also F. W. Taussig, International Economics (New York, 1928), 
Part I. 

52 C. F. Bastable, The Theory of International Trade, 4th ed. (London, 1903), pp. 


ratio of the marginal costs of the two products. 03 Haberler then demon- 
strated that the gains from international trade could be indicated, if not 
measured, by means of his production substitution curves. Leontief later 
made this idea more precise by combining the production substitution 
curve with a system of indifference curves. 54 

The principal advantages of the new approach, as Haberler pointed 
out, were, first, that it dispensed with the labor cost or real cost theory 
of value and, second, that it enabled one to consider a number of differ- 
ent factors of production simultaneously. Viner, however, argued that 
these advantages were illusory, and that the substitution curve concealed 
a number of important problems. In particular, he attacked the assump- 
tion of a fixed quantity of factors of production which is implied in the 
substitution curve. He insisted that the substitution curve could not be 
accepted as a fixed curve, determined by technological conditions, be- 
cause the amount of each of the factors of production was not fixed but 
depended upon its price. The latter, in turn, was influenced by inter- 
national trade. 55 In short, Viner argued that if the supply of the factors 
of production can be varied, the "real cost" of supplying such services 
must be taken into account along with the utility of commodities, in 
measuring the gains from trade. A similar objection was raised with 
regard to the use of indifference curves, but this is an older argument 
and is perhaps a less distinctive contribution of Viner than his discussion 
of real costs. The essence of the argument is that indifference curves for 
a country as a whole depend not only upon the amounts of the commodi- 
ties but upon their distribution between different individuals, and since 
international trade affects the distribution of income it also produces a 
shift in the community indifference curves. 

Considering these logical flaws in the theory of international trade, 
it is natural to inquire what remains of the concept of gains from inter- 
national trade. If the assumption of fixed quantities of resources is re- 
jected, and if it is impossible to derive meaningful indifference curves 
for a country as a whole, in what sense can we say that international trade 
contributes to welfare? Answers to these questions were given in 1939 
by Samuelson, who demonstrated that even without the restrictive classi- 
cal and neo-classical assumptions international trade involves a potential, 
if not an actual, economic gain to all participating countries. 56 Samuel- 

63 Gottfried Haberler, The Theory of International Trade (New York, 1936), pp. 

64 W. W. Leontief, "The Use of Indifference Curves in the Analysis of Foreign 
Trade," Quarterly journal of Economics, May 1933, XLVII, pp. 493-503. 

55 Jacob Viner, Studies in the Theory of International Trade, pp. 516-526. 

56 Paul A. Samuelson, "The Gains from International Trade," The Canadian journal 
of Economics and Political Science, May 1939, V, pp. 195-205. 


son's argument consisted essentially in proving that, if trade is opened 
between countries, each country, if it chooses to do so, can obtain more 
of every commodity while performing less of every productive service. 
Although no objective measure can be given for the total gain, in this 
case, it is nevertheless clear that an increase of the quantity of every 
commodity and a decrease in the amount of every type of work per- 
formed represents an improvement of welfare. In this sense, international 
trade involves a gain for all countries. Samuelson was careful to point out 
that his demonstration did not imply that completely unrestricted trade is 
the optimum position for all countries. His argument, on the contrary, 
was restricted to the proposition that some degree of trade, however re- 
stricted or unrestricted it may be, is necessarily better for all countries 
than no trade at all/' 7 The modern conception of the gains from trade 
thus provides no answer to the problem of free trade vs. protection. It 
simply shows that protection, if carried to the point where all trade is 
eliminated, will reduce welfare, compared with any intermediate posi- 


We have argued above that new developments in price theory have 
refined but have not fundamentally altered the classical concept of the 
gains from international trade. In one respect, however, the application 
of innovations in price theory to the special problems of international 
trade has been more revolutionary. Although the classical theory gave a 
good explanation of the gains which international trade brings to a coun- 
try as a whole, the traditional theory was never able to explain adequately 
how these gains are distributed between different factors of production 
or between different industrial groups. In other words, the classical 
theory could not explain the relation of international trade to the distri- 
bution of income. The clarification of this relation during the interwar 
period was therefore a major achievement in international economics. 

It can hardly be said that the classical economists were entirely un- 
aware of the influence which foreign trade can exert upon the distribu- 
tion of a country's income. The later development of the classical theory 
itself was to a considerable extent an outgrowth of the controversy in 
England over the Corn Laws, and one of the principal issues in this 
controversy was the conflict of interests between owners of agricultural 

67 In a reflective and perhaps nostalgic mood, an economist once remarked that the 
argument over whether some trade is necessarily better than no trade at all reminded 
him of a favorite cliche of his college days: "Prohibition is better than no liquor at all." 


land and industrialists. With the help of the Ricardian theory of rent, 
the English economists of the nineteenth century were able to show 
clearly that tariff reductions in a country (such as England) importing 
agricultural products could reduce the landowners' share of the national 
income. 58 But despite their insight into this particular problem, they 
never succeeded in integrating their international economics with a gen- 
eral theory of distribution. This is hardly surprising, for until Clark, 
Marshall, VVicksteed and others had generalized the Ricardian law ot 
diminishing returns into a law of variable proportions for all factors of 
production, no completely general theory of distribution was available. 5!> 
Even after the theory of distribution had been discovered there was a con- 
siderable lag in applying it to international problems, and the theory of 
distribution did not become a systematic part of international economics 
until the interwar period of the present century. 60 

The pioneer work was a Swedish essay of 191 9 by Heckscher. 61 Be- 
cause of language difficulties, this essay was generally neglected by 
English and American economists, but it is now recognized as one of the 
first contributions to an important development in international econom- 
ics. Heckscher's contribution consists essentially in explaining the flow 
of international trade in terms of the relative scarcity or abundance of 
different factors of production. -The classical economists had explained 
the flow of trade by means of their law of comparative advantage, but 
the fact of comparative advantage itself had been accepted more or less 
without explanation. By using the generalized theory of production— the 
doctrine of marginal productivity— Heckscher was able to show that in 
many instances the advantage which a country enjoys in the production 
of a particular commodity is attributable to a large supply, relative to 
other countries, of the factor or factors of production which are most 
important in that commodity. In other words, if Country A has more 
land per laborer than Country B, rents, relative to wages, will be lower 
in the former country than in the latter. Country A will therefore have a 

cs J. S. Mill, Principles of Political Economy, Book V, Ch. IV, Sec. 5. 
C9 George J. Stigler, Production and Distribution Theories (New York, 1941), passim, 
but especially Ch. I. 

60 In his Theory of International Trade (Dublin, 1887), C. F. Bastable had included 
a chapter on "The Influence of Foreign Trade on the Internal Distribution of Wealth," 
but his discussion did not go much beyond that of the classical economists. Cf. also 
Simon N. Patten, The Economic Basis of Protection (Philadelphia, 1890), Ch. V. 

61 Eli F. Heckscher, "Utrikhandelns verkan pa inkomstfordelningen," Ekonomisk 
Tidskrift, 1919, Del II, pp. 1-32. Among English-speaking economists, this work has 
been known largely through the writings of Bertil Ohlin. Fortunately, we are soon to 
have an English version of Heckscher's paper, translated from the Swedish by Professor 
and Mrs. Svend Laursen, in Readings in the Theory of International Trade, to be pub- 
lished under the auspices of the American Economic Association. 


comparative advantage in wheat and other products requiring much land 
and little labor, while Country B has a comparative advantage in manu- 
factures. Although this proposition may now seem self-evident, it never- 
theless represented a major improvement in the theory of international 
trade, for it opened the door to a systematic treatment of the relation 
between international trade and the distribution of income within a 
single country. Thus, by demonstrating how international trade increases 
the demand for the factors of production which a country has in relative 
abundance, while reducing the demand for its relatively scarce factors, 
Heckscher also showed that international trade has a tendency to equal- 
ize the relative returns to land, labor, and capital, throughout the world. 
The full significance of this conclusion will become apparent in our later 
discussion of the theory of tariffs. 

Heckscher's analysis concerning the relation of comparative advantage 
to relative quantities of the factors of production, and his conclusion that 
international trade has an equalizing tendency upon relative factor prices, 
later became the basis for the well-known treatise of Ohlin. 62 Ohlin modi- 
fied and refined the Heckscher theories in several respects. Among other 
things, he considered the possibility that a change in relative factor 
prices, brought about by international trade, might eventually alter the 
available supply of some of these factors. 63 He also took account of the 
complications which arise when more than two factors of production are 
considered, and when relations of complementarity exist among some of 
the factors. 64 Apart from such complications, however, Ohlin's conclu- 
sions were in substantial agreement with those of Heckscher. 


To complete our discussion of price theory and international trade, a 
word should be said about two studies in the field of general equilibrium. 
The classical theory, in its rigorous form, dealt only with the problem 
of two countries trading in two commodities, and subsequent revisions 
or refinements of the classical theory have seldom gone beyond this 
simple framework. Even the work of Ohlin, which is sometimes consid- 
ered as a more general approach, is best understood in the classical con- 
text; i.e., the Heckscher-Ohlin conclusions can be most clearly stated 
and most rigorously demonstrated if the theoretical scheme is limited 
to two countries, two commodities, and two factors of production. The 
only rigorous attempts, so far as I am aware, to develop a completely gen- 

62 Bertil Ohlin, Interregional and International Trade (Cambridge, Mass., 1933). 

63 Ibid., Ch. VII. 

84 Ibid., pp. 97-99. 


eral approach to international economics, are the studies by Yntema 6 '' 
and Mosak. 66 Yntema's work appeared in 1932, and the work of Mosak, 
which the latter regards as a sequel to Yntema's book, was published 
twelve years later. 

Except for their greater degree of generality, both of these books 
adopted a distinctly classical point of view, and in the main their conclu- 
sions confirmed the classical reasoning. Like the classical economists, 
Yntema and Mosak were interested primarily in the effects of interna- 
tional disturbances, such as tariffs and indemnity payments, upon rela- 
tive prices at home and abroad. Apart from a final chapter in Mosaks 
book, little attention was paid in either book to the influence of inter- 
national trade on the level of output. In at least one respect, however, 
both books have gone considerably beyond the classical theory: both 
authors have shown that a study of dynamic economics, of stable and 
unstable market systems, can be extremely useful in the field of inter- 
national trade, particularly in complex problems involving a large num- 
ber of countries and a large number of commodities. 

Yntema's study of dynamic problems was necessarily elementary, since 
he was writing at a time when little was known about the stability of 
market systems. Nevertheless, he carried the discussion of stable and 
unstable international markets beyond the pioneer stage at which Map 
shall had left it, 67 and in Chapter V of his book he adopted a method 
which was surprisingly similar to the dynamic approach which Hicks 
later presented in Value and Capital. He argued, for example, that a 
money payment from Country 2 to Country 1 would normally increase 
money prices and costs in the latter and reduce them in the former, 
essentially through the operation of the quantity theory of money. In 
studying the static equations of supply and demand, he therefore sug- 
gested that any shapes of these functions, such as extremely inelastic 
demands, which indicated that a money payment would reduce prices 
in the receiving country and raise them in the paying country could be 
rejected as fundamentally unstable. 68 Here, in a book published in 1931, 
is the germ of the Hicksian concept of imperfect stability. Elementarv 
and incomplete as it was, Yntema's conception of the relation between 
dynamics and statics was nevertheless prophetic of the work to be done 
later in the field of general price theory. 

c " Theodore O. Yntema, A Mathematical Reformulation of the General Theory of 
International Trade (Chicago, 1932). 

06 Jacob L. Mosak, General-Equilibrium Theory in International Trade (Bloomington, 

m Cf. Alfred Marshall, op. cit., Appendix J. 

68 Theodore O. Yntema, op. cit., p. 80. 


Mosak began his study of general equilibrium where Yntema had 
ended, and in several respects he advanced the theory of international 
trade beyond the stage at which Yntema had left it. Perhaps most impor- 
tant, Mosak took account of the effects of shifts in purchasing power, 
such as those associated with indemnity payments, on the demand for 
internationally traded goods. Yntema had neglected such income effects, 
and in this respect his system was even more classical than that of many 
of the classical economists. Except for Chapter IX of his book, Mosak 
dealt with a system in which incomes were entirely expended on com- 
modities, either foreign or domestic. For this reason, the "balance-of- 
payments problem," as such, did not enter explicitly into most of his 
work; equality between supply and demand in all commodity markets 
in Mosak's system, implies equilibrium in the foreign exchange markets. 
He made a sweeping application of Hicks' concepts of perfect and imper- 
fect stability to the special problems of international trade, but his con- 
clusions in. this regard are vitiated to some extent by the fact that the 
Hicks conditions are not true stability conditions except under special cir- 
cumstances. 69 

It is extremely difficult, in a review such as this, to evaluate the position 
or importance of the two general-equilibrium studies in the body of inter- 
national economics as a whole. Unlike the classical theory, the more 
general approaches to international economics have not had a profound 
influence upon economic policy. To some extent this is probably attribut- 
able to their complexity— to the fact that the general solutions admit 
many different possible consequences of a given policy. To some extent, 
also, it may be a result of the fact that the more general theories have not, 
on the whole, revealed any serious flaws in the classical position. Both 
Yntema and Mosak discuss the classical theories as special cases of their 
more general theories, and their results tend largely to confirm the classi- 
cal reasoning. 70 

V. The Theory of Tariffs 

The preceding account of the relations between price theory and inter- 
national economics has considered only the logical development of the 
theory, and no attempt has been made to relate this development to prob- 
lems of economic policy. For this reason, the discussion above probably 
has an unrealistic and abstract quality which belies the true nature of the 

w See Paul A. Samuelson, ''The Stability of Equilibrium : Comparative Statics and 
Dynamics/' Econometrica, April 194 1, IX, pp. 97-120; and Lloyd A. Metzler, "Stability 
of Multiple Markets: the Hicks Conditions," ibid., October 1945, XIII, pp. 277-292. 

70 Theodore O. Yntema, op. cit., pp. 80-87; Jacob L. Mosak, op. cit., Ch. IV. 


classical tradition. In the actual course of its growth, the accepted theory 
of international trade was intimately connected with highly controversial 
questions of policy. Indeed, almost all of the major contributions to the 
theory of international trade were a direct outgrowth of practical eco- 
nomic issues. During the nineteenth century, foreign trade policy was for 
the most part synonymous with tariff policy, and the theory of tariffs ae 
cordingly occupies a prominent place in the classical economics. Even 
during the interwar period of the twentieth century, when other methods 
of trade control such as import quotas, exchange controls, etc., were 
widely adopted, tariffs continued to occupy an important place in the 
commercial policies of many countries. It seems appropriate, therefore, to 
conclude this review with a brief discussion of some recent changes in 
the theory of tariffs. 

The tariff literature is perhaps as voluminous as that of any branch of 
applied economics, but fortunately it will not be necessary to make a de- 
tailed study of all the books, articles, and pamphlets dealing with this 
subject. Much of the writing about tariffs is concerned with problems of 
politics, ethics, or administration, and all of these questions, however im- 
portant they may be, are outside the scope of the present review. With 
regard to the purely economic issues, recent developments in the theory 
of tariffs may be grouped under three main heads: (i) tariffs and the 
terms of trade; (2) tariffs and the distribution of income; (3) commercial 
policy and the revival of mercantilism. Each of these will be considered 
in turn. 


Events of the interwar period led to a notable revival of interest in 
tariffs regarded as bargaining weapons. The Reciprocal Trade Agree- 
ments program in the United States, the new tariff policy of the United 
Kingdom, and tariff increases in other countries all emphasized once 
more the monopolistic character of tariff restrictions, and the effects of 
tariffs on the terms of trade. A number of economists dealt with these 
problems, but perhaps the outstanding contributions were those of Sam- 
uelson, 71 Kaldor, 72 Benham, 73 and Scitovszky. 71 The outcome of the dis- 
cussion was to demonstrate, anew, the shaky foundation of some of the 
arguments for free trade. Samuelson, for example, pointed out that tariffs 

71 Paul A. Samuelson, "Welfare Economics and International Trade," American Eco- 
nomic Review, June 1938, XXVIII, pp. 261-266. 

7 ~ Nicholas Kaldor, "A Note on Tariffs and the Terms of Trade," Economica, Novem- 
ber 1940, New Series VII, pp. 377-380. 

7a Frederic Benham, "The Terms of Trade," ibid., pp. 360-376. 

7 * Tibor de Scitovszky, "A Reconsideration of the Theory of Tariffs," Review of 
Economic Studies, Summer 1942, IX, pp. 89-110. 


or other trade restrictions, if they improve the terms of trade and increase 
the welfare of the country imposing them, will do so by worsening the 
position of some other country. Since the gains of one country cannot be 
measured against the losses to another, there is no presumption that trade 
restrictions always reduce welfare for the world as a whole. Scitovszky 
carried this point of view further by showing that a tariff in one country 
increases the probability that retaliation will prove profitable to other 
countries. He thus developed a theory of tariff retaliation which depicted 
each country as raising tariffs in a rational manner in order to secure for 
itself a more favorable position in world markets. Although each country 
separately might gain, in Scitovszky's view, by a moderate increase in 
tariffs, the effect of all countries following the same policy would be to 
reduce the welfare of each of them. In the end, therefore, Scitovszky be- 
lieved tariff retaliation would lead either to bilateral barter deals or to 
some form of tariff bargaining. Since each country may have a rational 
interest in higher tariffs, quite apart from the pressure of special groups 
within the country, Scitovszky emphasized that a free-trade system, like 
a cartel, has a natural tendency to disintegrate, and must be enforced by 
some kind of international convention. 

The general effect of the work by Samuelson, Scitovszky, and others 
was to call attention to the similarity between the theory of tariffs and the 
theory of monopoly. An individual country, under certain conditions, can 
gain by limiting trade with other countries, just as a monopolist can gain 
by restricting the supply of the monopolized product. The rational argu- 
ment against tariffs is not, as free-traders sometimes suppose, that tariffs 
harm all countries, but that, like monopoly restrictions, they impose a 
loss on some countries which is greater than the gain to the country im- 
posing them. In modern terminology, we would say, following Lerner, 
that tariffs result in an inefficient allocation of resources, since the price 
ratios between commodities differ from one country to another. 75 


What relation does this recent work have to the general literature on 
the theory of tariffs? Perhaps most important, the new contributions to 
tariff theory correct a misconception which has long been common among 
advocates of free trade. In popular discussions, and even among some 
professional economists, the classical theory of comparative advantage 
has frequently been presented as a proof that all countries gain, individu- 
ally, by a policy of unrestricted commerce. When the monopolistic char- 
acter of tariffs and other impediments to trade is considered, however, it 

~~" Abba P. Lerner, The Economics of Control (New York, 1944), pp. 356-362. 


is obvious that this popular view is incorrect. Tariffs do not always involve 
a decline in the welfare of all countries, but may involve a loss in some 
countries and a gain in others. The importance of this misconception re- 
garding tariffs, and the tendency to disregard the influence of tariffs on 
the terms of trade, are well illustrated by the following quotations from a 
report on tariffs prepared in 1930 by a distinguished committee of British 

As a matter of history, the assertion that the advantages of Free Trade depend 
upon its being mutual, has always been made by people who were attacking Free 
Trade. It has never been made by any of the principal advocates of Free Trade. 
For this there is a simple reason. It represents complete misunderstanding of the 
nature of international trade and the working of tariffs. 

International trade is never free of all obstacles. The argument of the Free 
Traders has been directed to making the obstacles as few as possible. The gain 
through removing one obstacle depends in no way at all upon the removal of all 
the other obstacles or any of them. 

. . . For other countries to tax our exports to them is an injury to us and an 
obstacle to trade. For us to tax their exports to us is not a correction of that injury; 
it is just a separate additional obstacle to trade. 76 

In the light of recent tariff discussions, statements such as these obvi- 
ously need to be made more cautiously and with more reservations. As a 
practical matter it may well be true that actual tariffs have been so high 
as to inflict injury upon all parties, but there is no presumption, from the 
law of comparative advantage, that this will always be true. The argu- 
ments for retaliation are stronger than the literature on free trade would 
lead one to suppose. 

Although the recent contributions to the theory of tariffs have made an 
important correction in free trade arguments such as those above, it 
would be a mistake to suppose that the modern view represents a pro- 
nounced divergence from the classical theory. The English economists of 
the nineteenth century, while they did not perhaps give the point suffi- 
cient emphasis, were fully aware of the favorable effects which a tariff 
may have upon a country's terms of trade. 77 Indeed, one of the primary 
purposes of Mill's profound work in international trade was to show how 
tariffs and other obstacles to trade affect the ratio of interchange between 
exports and imports. As Taussig has shown, 7S Mill probably drew tOo 
sharp a distinction between revenue duties and protective duties, but 

76 Sir William Beveridge et al., Tariffs: The Case Examined (London, 1931), pp. 
108-1 10. 

77 Cf . Marion C. Samuelson, "The Australian Case for Protection Reexamined," Quar- 
terly Journal of Economics, November 1939, LIV, pp. 147-148. 

78 F. W. Taussig, International Trade, p. 146. 


with regard to revenue duties he, Mill, never had any misconceptions 
about the benefits which a single country could derive, at the expense of 
its neighbors, from imposing such duties. The following quotation, if 
any is necessary, should make this obvious: 

A country cannot be expected to renounce the power of taxing foreigners, unless 
foreigners will in return practice toward itself [sic] the same forbearance. The only 
mode in which a country can save itself from being a loser by the duties imposed 
by other countries on its commodities, is to impose corresponding duties on theirs. 79 

Marshall, 80 Edgeworth, 81 and Taussig 82 were likewise untouched by 
the error of the free-traders. All of these economists recognized the possi- 
ble gains which a country may achieve by means of tariffs. Marshall, 
however, felt that as a practical matter the terms-of-trade argument for 
tariffs was not of great importance. In the first place, he doubted whether 
any single country in the modern industrial world was of sufficient im- 
portance to have an appreciable effect on the terms of trade. Second, his 
observations, particularly in the United States, led him to the conclusion 
that the pressure of special interests usually forces tariff rates far above 
the optimal level, so that the country imposing such tariffs, as well as the 
rest of the world, is likely to lose. 83 

Space limitations forbid any further elaboration of this theme. The 
point to be emphasized is the continuity between the classical theory and 
the recent discussions of the relation of tariff to the gains from trade. By 
utilizing modern price theory, contemporary economists have given a 
more precise statement of the possible gains and losses from tariffs, and 
they have indicated, perhaps more accurately than before, the limits to a 
rational tariff policy; but they have not altered the fundamental princi- 
ples as developed by Mill, Marshall, Edgeworth, and Taussig. Further 
elaboration was needed, not because the classical view was incorrect, but 
because it was not given sufficient emphasis, and, as a consequence, was 
later misinterpreted by advocates of free trade. 


The preceding discussion has been concerned primarily with the 
effects of tariffs on the distribution of world income between countries. 

* 79 John Stuart Mill, Essays on Some Unsettled Questions of Political Economy, p. 29. 

80 Alfred Marshall, Money, Credit and Commerce, Appendix J and Ch. VIII-X. 

81 F. Y. Edgeworth, op. cit., Vol. II", Sec. IV. 

83 F. W. Taussig, International Trade, Ch. XIII. 

83 It is interesting to note, also, the similar remark of Edgeworth on this point: 
". . . protection might procure economic advantage in certain cases, if there was a 
Government wise enough to discriminate those cases, and strong enough to confine itself 
to them; but this condition is very unlikely to be fulfilled." Edgeworth, op. cit., Vol, II, 
p. 18. 


An equally important problem, and one which was never satisfactorily 
solved by the classical economists, concerns the effects of tariffs upon the 
distribution of income within a single country. Even though a country 
has no monopolistic position in world markets and cannot gain a larger 
share of the world's income by restricting trade, it may nevertheless be 
true that a particular class or group within the country would be bene- 
fited by a tariff. This question of income distribution has been a central 
issue in almost all tariff controversies. In the United States, it took the 
form of an assertion that tariffs on manufactures protect or raise the real 
wages of workers. In the United Kingdom the issue appeared in the 
familiar conflict between landlords and manufacturers over the effects of 
the Corn Laws on agricultural rents and real wages. More recently, the 
question arose again in a report on the Australian tariff which was pre- 
pared in 1929 by a distinguished group of Australian economists. 84 The 
outstanding feature of this report was that it gave qualified support to the 
tariff as a permanent part of Australia's commercial policy. Although the 
terms-of-trade argument played some part in the conclusion of the Aus- 
tralian economists, their principal reason for favoring protection was the 
belief that tariffs maintained a better distribution of income between 
landlords and workers than would otherwise have been possible. A tariff 
reduction, according to the report, would reduce the Australian output of 
manufactured goods and increase the output of the principal export 
goods, wool and wheat. And since wages are normally a much greater 
proportion of the value of manufactures than of the value of agricultural 
products, the effects of this shift in the composition of Australian produc- 
tion would be to increase the demand for land and reduce the demand 
for labor. Workers would accordingly receive a smaller proportion of the 
national income than under protection, even if labor were perfectly 
mobile between agriculture and industry. 85 

The Australian report, as well as the tariff controversies in the United 
States, the United Kingdom, and other countries, have all emphasized 
the influence of tariffs on the distribution of income. In this particular 
branch of economic policy, however, economic theory, until recently, has 
had little to contribute. As noted earlier, the classical economists, with 
their over-simplified theory of production, were unable to cope with the 

81 The Australian Tariff, an Economic Enquiry, a report prepared for the Common- 
wealth Government by J. B. Brigden, D. B. Copland, E. C. Dyason, L. F. Giblin, and 
C. H. Wickens (Melbourne, 1929). 

85 Thus, on p. 5 the report states: "The tariff has had the effect of pooling the national 
income to a greater extent than would have been practicable it assistance to industry were 
derived solely through the more obvious method of taxation. Employment has been subsi- 
dized at the expense of land values, enabling the standard of living to be maintained with 
a rapidly increasing population." 


problem. 86 They realized, of course, that protection benefits the producers 
of particular commodities, and that factors of production which are spe- 
cific to these protected industries will likewise gain from tariffs. But if 
labor and capital are mobile, and the standard wage rate tends to equality 
in all industries, the traditional theory of international economics cannot 
show how tariffs affect the distribution of income between such broad 
categories as "labor," "capital," and "land." In the absence of a theory of 
production which recognized the substitution of one factor for another, 
there was a tendency to assume that the reward of each of the factors 
depended upon the productivity of the entire economy, and to conclude 
that foreign trade, which increases the effectiveness of the economy as a 
whole, must also increase the rewards to each of the factors separately. 87 
Although widely held, this view was by no means universal even 
among followers of the classical tradition. As early as 1906, for example, 
Pigou had pointed out that a tariff increases the output of one industry, 
A, at the expense of another industry, B, and that if one factor of pro- 
duction plays a more important part in A than in B, the change in the 
composition of the national income will increase the proportion of the 
total product accruing to that factor. 88 Moreover, even if the tariff reduces 
the national real income as a whole, Pigou argued that the absolute posi- 
tion of the favored factor might be improved. "The increase per cent in 
the share of the dividend obtained by the favored factor might exceed the 
shrinkage per cent of the dividend itself." 89 Viner and Ohlin later ex- 
pressed similar views, although the latter regarded it as unlikely that the 
absolute as well as the relative returns to the favored factor would be 
increased by protection. 90 Thus during the interwar years there seems to 
have been agreement that tariffs tend to increase the relative share of the 
national income accruing to certain factors of production, but there were 
doubts as to whether the absolute returns to the favored factors are also 
increased. If protection went so far as to reduce real income as a whole, it 
was. believed that the absolute returns to the favored factors might con- 
ceivably be reduced even though their relative share in total income was 
increased. In other words, economists were still uncertain about the 
effects of protection upon the real income of certain factors, even though 

88 This is perhaps a slight exaggeration, for the classical economists had a great deal to 
say about the effects of the Corn Laws on agricultural rents. See, e.g., three letters of 
Colonel Robert Torrens to the Marquis of Chandos (London, 1839). 

87 See, e.g., F. W. Taussig, "How the Tariff Affects Wages," in Free Trade, the Tariff 
and Reciprocity (New York, 1920), p. 59. 

88 A. C. Pigou, Protective and Preferential Import Duties (London, 1906), p. 58. 

89 Ibid., p. 59. 

w Jacob Viner, Studies in the Theory of International Trade, pp. 533-534; Bertii 
Ohlin, Interregional and International Trade, p. 44. 


they were agreed that such protection would probably increase the fro- 
fortion of the national income accruing to the favored factors of pro- 

Much of this uncertainty was eliminated in 1941 in a paper by 
Stolper and Samuelson which presented a remarkable application of the 
Heckscher-Ohlin system to the special problem of tariffs. 91 The paper 
demonstrated that, regardless of its effects on the term of trade and real 
income as a whole, protection, unless followed by retaliation, always in- 
creases the real return as well as the relative share in the total product of 
the factor of production which is relatively most important in the pro- 
tected industries. Suppose, for example, that a country has two industries, 
clothing and food, and two factors of production, labor and capital. Sup- 
pose further that labor is more important in the production of clothing 
than in the food industry, and that the country is an importer of clothing. 
A tariff on clothing will then increase production in that industry and 
shift both capital and labor from agriculture to manufacturing. Accord- 
ing to the Stolper-Samuelson argument, this shift of resources will neces- 
sarily reduce the proportion of labor to capital in both industries. In other 
words, the food industry, where a comparatively small amount of labor 
is employed, cannot supply enough labor to maintain the old labor-capital 
ratio in the clothing industry without creating a disparity between the 
marginal productivities, and consequently the wage rates, in the two 
industries. Now if the ratio of labor to capital is reduced in both indus- 
tries, it follows from the law of variable proportions that the marginal 
product of labor, and hence the real wage rate, must be higher than 
before, regardless of whether real wages are measured in terms of the 
protected commodity, clothing, or the unsheltered commodity, food. 
Thus if protection increases the money price of clothing, it must increase 
the money wage rate even more. The real return to the second factor, 
capital, is reduced, compared with the free-trade position, for the shift in 
the composition of output brought about by the tariff makes capital a 
more abundant factor in both industries. And since the quantities of the 
factors are assumed to be unaffected by the tariff, it is clear that the tariff 
has increased the relative as well as the absolute returns to labor, and has 
reduced both the relative and absolute returns to capital. This conclusion, 
which Stolper and Samuelson reached in their paper, does not depend 
upon any monopolistic or other restrictions to the movement of factors of 
production. On the contrary, it assumes that labor and capital receive the 
same return in all industries. Neither does the result depend upon a gain 

91 Wolfgang F. Stolper and Paul A. Samuelson, "Protection and Real Wages," Review 
of Economic Studies, November 1941, IX, pp. 58-73. 


in real income for the country as a whole, such as that which might arise 
from a favorable movement in the terms of trade. Even when protection 
leaves the external prices of exports and imports unchanged, and when 
the national real income as a whole is accordingly reduced, it remains 
true that the real income of the factor most important in the protected 
industries is increased by the tariff. 

When we attempt to make practical applications of the Stolper-Sam- 
uelson tariff argument, a number of complications arise. Perhaps the most 
important of these is the large number of factors of production. The con- 
clusion summarized above concerning the effects of tariffs on the distri- 
bution of income can be rigorously proved only for the simplified case of 
two factors of production; when more than two factors are involved, the 
terms "relatively scarce factors" and "relatively abundant factors" lose 
some of their precise meaning. Moreover, if some of the factors are com- 
plementary it is no longer possible to say that the marginal product of a 
particular factor depends exclusively upon its amount, relative to some 
other factor. 92 Despite this complication, however, it seems reasonable, as 
Stolper and Samuelson have suggested, 93 to propose a number of tenta- 
tive conclusions. In the United States during the early nineteenth cen- 
tury, for example, our comparative advantage in agriculture was clearly 
governed by the large amount of land per worker. Under these circum- 
stances, the tariffs which were imposed upon manufactured imports may 
well have improved the standard of living of the working class as a whole, 
since labor is a more important factor in manufactures than in agricul- 
ture. The same argument is applicable to the present position of Aus- 
tralia. In other words, the Stolper-Samuelson theorem provides a scientific 
foundation for the conclusion of the Australian committee that the tariff 
has helped to maintain the living standard of the working class. 

Recent discussions of tariffs and the distribution of income thus give 
limited support to the "pauper labor" argument for tariffs. We would no 
doubt agree today with Taussig that "no economist of standing would 
maintain that a protective tariff is the one decisive factor in making a 
country's rate of wages high," but it is doubtful whether we would also 
agree that "no economist, . . . would sanction the pauper-labor argu- 
ment for tariffs." On the other hand, as Viner has pointed out, the favor- 
able effect on the distribution of income from the point of view of one 
factor of production is not, if so facto, a valid argument for tariffs from 
the point of view of the country as a whole. If repercussions on the terms 
of trade can be neglected, tariffs always reduce the total quantity of com- 

02 See ibid., pp. 72-73. 

03 Ibid., p. 73. 


modities available to the country as a whole, and the gain to one factor of 
production is therefore more than offset by losses to others. 91 


Having touched upon two aspects of the theory of tariffs, it seems ap- 
propriate to conclude this section with a few remarks concerning com- 
mercial policy as a whole. Although this subject is much broader than the 
theory of tariffs, the tariff question has been at the center of so many of 
the historical conflicts in commercial policy that it seems best to consider 
the broader subject here. 

The basic conflicts in the field of commercial policy are well illustrated 
by the familiar differences of opinion between the mercantilists and the 
classical economists. # To the mercantilists, the primary functions of for- 
eign trade were to provide an outlet for a country's surplus production 
and to acquire a large stock of the monetary metals. 95 * In order to achieve 
these ends, they advocated tariffs, export subsidies, and other measures 
designed to assure a country a steady and substantial export surplus./The 
classical system of economic theory was developed, in part, as a refutation 
of these mercantilist doctrines, and although the classical theory has been 
misinterpreted and misused by the advocates of free trade, it remains 
nevertheless as a bulwark against protectionism and other commercial 
restrictions. «The fundamental tenet of the classical theory was that the 
purpose of all economic activity is not to find markets for surplus produc- 
tion or to increase mercantile profits but to satisfy human wants. ^Foreign 
trade, like other economic activity, was to be evaluated according to its 
contribution to this want-satisfying function. 3The classical economists ac- 
cepted Say's Law and therefore believed that general over-production was 
impossible. From this it followed that the primary problem in economics 
was not how to find employment for unused capital, labor, and land, but 
how to use these resources in the most effective manners 

In the field of international trade, this meant a shift in emphasis from 
exports to imports. The usefulness of exports was to be judged not by 
their contribution to output and employment but by the value of the im- 
ports obtained in exchange. To the mercantilist argument that export 
subsidies were desirable because they created an outlet for surplus pro- 
duction, the classical economists replied that no such outlet was needed, 
and that an increase of exports, in itself, meant a smaller amount of goods 
available for domestic consumption. 90 In other words, exports were con- 

04 Jacob Viner, Studies in the Theory of International Trade, pp. 533-534. 

95 Cf. James VV. Angell, The Theory of International Prices, Ch. II. 

90 Ct. John Stuart Mill, Principles of Political Economy, Book V, Ch. X, Sec. 1. 


sidercd to be desirable only if the additional imports which could thereby 
be obtained were worth more to consumers than the home consumption 
which had to be sacrificed. From this it followed that an export surplus, 
financed by imports of the precious metals, was not a desirable long-run 
policy, since monetary stocks of metal have no intrinsic ability to satisfy 
wants. It would be better, according to the classical view, to export only 
enough to pay for imports and to use the additional resources to produce 
for home consumption. In any event, the classical system envisaged an 
automatic tendency toward a balancing of exports and imports through 
the effects of gold movements on the level of prices. Attempts to main- 
tain an export surplus were therefore regarded as self-defeating: tariffs 
and other impediments to trade, instead of increasing employment at 
home, would simply transfer resources from export industries to domestic- 
goods industries, thereby reducing the effectiveness of labor and other 
factors in satisfying wants. 

Among economists in England, this classical system rapidly supplanted 
the mercantilist doctrines, and it was by far the most widely accepted 
system throughout the nineteenth century. Among businessmen and 
statesmen, on the other hand, the mercantilist doctrines continued to 
have much influence. Time and again, the argument that tariffs increase 
employment appeared in popular discussions, and just as regularly these 
arguments were refuted— sometimes impatiently— by economists. 97 Apart 
from the appeal to special interests which is no doubt the dominant factor 
in almost all tariff legislation, the genuine economic reasons for the per- 
sistence of the mercantilist arguments did not become apparent until the 
decade of the nineteen-thirties when the Keynesian revolution led to a 
reconsideration of the classical position. It was Keynes himself who 
pointed out the grain of truth in the mercantilist system, and except for 
the fact that this review is supposed to cover the whole field of interna- 
tional economics it would perhaps be best simply to refer the reader to 
Keynes' discussion. 98 The classical argument against an export surplus as 
a permanent policy stands or falls with the acceptance or rejection of the 
idea that the economic system tends automatically toward a state of full 
employment, and since Keynes had rejected this idea he was bound to 
reject also the idea that encouragement of exports through subsidies or 
reduction of imports through tariffs will have no influence on employ- 
ment. If a country starts from a position in which it has unemployed re- 
sources, it is no longer true, as the classical economists assumed, that an 

97 See, e.g., F. W. Taussig, presidential address before the American Economic Associ- 
ation, 1904, reprinted in Free Trade, the Tariff and Reci'procity , p. 29. 

98 J. M. Keynes, General Theory of Employment, Interest and Money (New York, 
1936), Ch. XXIII. 


increase of exports means a reduction of goods available for domestic 
consumption. On the contrary, if we consider the repercussions of higher 
income earned in the export trades on the demand for goods and services, 
it is probable that an increase of exports means also an increased demand 
for and an increased output of domestic goods. An export surplus 
financed by an inflow of gold may therefore be a direct cause of increased 
income and a higher standard of living in the exporting country. Thus 
the mercantilists may have been justified, in certain circumstances, when 
they advocated an export surplus as an outlet for excess production. The 
theoretical foundations of the classical view, as Keynes shows, were 
weaker than had been suspected. 

One problem still remains, however: How can the promotion of an 
export surplus as a permanent policy be reconciled with the classical ar- 
gument that imports tend automatically to balance exports? If a country 
reduces its imports by means of tariffs, will not the increased employ- 
ment in the protected industries soon be offset by a corresponding reduc- 
tion of exports? In order to answer these questions, we must refer again 
to what was said in Section II concerning the balancing mechanism* It 
must be conceded that the imposition of a tariff or the granting of a 
subsidy will set up a balancing process, although the mechanism is some- 
what different from that envisaged by the classical economists. If the 
initial disturbance is a tariff, imports will eventually rise, despite the 
tariff, as a result of higher income and employment at home, while ex- 
ports will decline as a result of lower income abroad. It was argued in 
Section II, however, that this balancing process is likely to be incom- 
plete. The modern theory of the balance of payments thus suggests that 
a country may have an export surplus over a considerable period of time 
without any pronounced automatic tendency toward a complete equaliza- 
tion of exports and imports. 

A large export surplus cannot be maintained indefinitely, however, for 
eventually the rest of the world will be drained of all its monetary gold. 
Even before this occurs, other countries will probably be provoked to re- 
taliation, particularly if unemployment prevails in the rest of the world. 
The most any individual country can hope to do is to maintain an export 
surplus which will ensure it a reasonable proportion of the world's new 
production of gold. If this policy succeeds, and retaliation is avoided, the 
export surplus, as Keynes shows, is doubly beneficial to the exporting 
country. On the one hand, it increases income and employment directly, 
and on the other hand, the increase of gold stocks, by reducing the rate 
of interest, tends to increase domestic investment. Both of these favor- 
able effects would have been denied by the classical economists, who saw 


no need for a stimulation of employment and who had a somewhat exag- 
gerated idea of the effectiveness of the balancing process. 

While all of this clearly indicates that the economic grounds for mer- 
cantilism were stronger than economists have generally recognized, the 
revival of the theory of mercantilism is not necessarily a reason for its 
advocacy as a practical policy. On questions of policy, the primary prob- 
lem, of course, is the reconciling of divergent national interests. The 
classical economists minimized this problem, for they believed that, 
within limits, national self-interest coincides with the welfare of the 
world economy as a whole. The revival of mercantilism demonstrates, 
unfortunately, that such harmony of interests cannot be taken for 
granted. The practical conduct of international trade is thus much more 
a problem of negotiation and compromise than the classical economists 
believed. If unemployment prevails through the world, as it did in the 
decade of the 'thirties, a mercantilist policy clearly benefits some coun- 
tries at the expense of others. And this easily leads to retaliation which 
deprives all countries of the benefits of international specialization with- 
out increasing employment in any of them. But on these political ques- 
tions, I can perhaps not do better than to quote a passage from Keynes* 
discussion of mercantilism: 

. . . There are strong presumptions of a general character against trade restrictions 
unless they can be justified on special grounds. The advantages of the international 
division of labour are real and substantial, even though the classical school greatly 
overstressed them. The fact that the advantage which our own country gains from 
a favourable balance is liable to involve an equal disadvantage to some other coun- 
try (a point to which the mercantilists were fully alive) means not only that great 
moderation is necessary, so that a country secures for itself no larger a share of the 
stock of precious metals than is fair and reasonable, but also that an immoderate 
policy may lead to a senseless international competition for a favourable balance 
which injures all alike. And finally, a policy of trade restrictions is a treacherous 
instrument even for the attainment of its ostensible object, since private interest, 
administrative incompetence, and the intrinsic difficulty of the task may divert it 
into producing results directly opposite to those intended." 

VI. Conclusions 

The review presented above of international economics during the in- 
terwar years has covered a wide variety of subjects. In the monetary part 
of the field, it has attempted to describe recent developments in the 
theory of the balance of payments under conditions of both fixed and 
flexible exchange rates. And in the so-called "pure" theory of interna- 

99 Ibid., pp. 338-339. 


tional trade it has presented an account of the adaptation of new discover- 
ies in price theory to the special problems of international trade. Finally, 
the recent changes in the theory of international trade have been related 
to selected problems in tariff policy. 

Considering the diversity of subjects discussed, it is perhaps useful to 
inquire, in conclusion, whether the recent innovations and discoveries in 
all the various branches of international economics have anything in 
common. Is there, for example, any unifying principle or any basic phi- 
losophy which unites the modern theories of the balance of payments 
with the revised theory of tariffs? Or has each of the new discoveries in 
each of the separate branches been a more or less isolated phenomenon? 
A cursory glance at the preceding pages may suggest that the latter has 
been true. I believe, however, that a more careful study will convince the 
reader that the new theories have not been as disconnected and isolated 
as seems at first to be the case. The connecting idea, however, is essen- 
tially negative. Historically, the interwar period will probably be remem- 
bered as a period of retreat from the price system, when all sorts of 
temporary or provisional measures were adopted to regulate economic 
activity. The market mechanism had broken down and no one seemed to 
know quite w 7 hy or just what to do about it. This was perhaps even more 
true of the international mechanism than of domestic markets, and to a 
very great extent the theoretical developments reflected the empirical. 
Where the classical economists had discussed the broad operation of the 
price system, twentieth-century economists described the exceptions and 
qualifications, or the special circumstances in which the international 
price mechanism would not work. Thus, for example, the balance of 
payments mechanism under the gold standard was found to be less effec- 
tive and more disruptive than the classical economists had believed. And 
even with flexible exchange rates, it was realized during the interwar 
years that a balancing of foreign receipts and expenditures cannot be 
taken for granted. Doubts concerning the price system were by no means 
limited to the monetary aspects of international economics. In the field of 
price theory, too, there was a movement away from traditional ideas. In- 
creasing emphasis was placed upon the fact that an unimpeded working 
of the free market system is not necessarily in the interest of each indi- 
vidual country. The classical conception of a harmony of interest be- 
tween countries, which even the classical theory did not entirely support, 
was called further in question. Much importance— perhaps too much- 
was attached to the benefits which individual countries could derive by 
the regulation of exports and imports. 

Part of this general retreat from the price system in international eco- 


nomic theory was no doubt beneficial, for it contributed to a more realistic 
appraisal of international trade than we had inherited from the classical 
economists. It seems likely, however, that the pendulum has now swung 
too far in the anti-classical direction. The interwar years give a distorted 
picture of the normal working of an international market system— indeed, 
it is more of a caricature than a picture— and economic theory has shared 
in this distortion. If in the past we have expected the price system to 
accomplish too much, there is a danger that in the future we shall expect 
it to do less than it is capable of doing. 

Our major error in the past, and the error which contributed perhaps 
more than anything else to discrediting of the price system, consisted in 
expecting the price mechanism to solve the problem of economic stability. 
On this point we were immensely enlightened during the interwar years, 
and it is now generally recognized that economic stability requires con- 
stant supervision and planning. To the extent that the world succeeds in 
solving the problem of stability by measures supplementary to the price 
system, the equally important problem of the allocation of resources, both 
domestic and international, might reasonably be left to the market mech- 
anism. In the absence of severe depressions, there is reason to believe that 
the balances of payments of most countries could be kept in reasonable 
equilibrium by means of moderate adjustments in exchange rates. The 
trade restrictions and trade controls which grew so rapidly in the inter- 
war years would therefore be unnecessary. Moreover, in a stable and ex- 
panding world economy, individual countries would have less incentive 
to adopt trade controls in order to safeguard their domestic markets. 
World economic stability would thus greatly reduce the force of the two 
most important incentives for controlling international trade in the past. 
Whether this would be sufficient to counteract the present trend toward 
increasing state intervention can hardly be foretold. But at any rate it is 
clear that our hopes for a revival of the market mechanism, however weak 
they may be, are greatly dependent upon a world stability which must be 
achieved, for the most part, by conscious planning and direction. 


Lloyd G. Reynolds 

Labor economics, like other parts of economics, consists of two principal 
kinds of work— the development of general propositions about the role of 
labor in the economy, and the testing of these propositions through fac- 
tual studies. Many of the central issues are not economic in the technical 
sense of being amenable to the categories of economic theory. A success- 
ful attack on them requires the diverse techniques of psychology, sociol- 
ogy, politics, law, and administration. The student of labor who wishes to 
grasp the phenomena of his field in toto must turn himself into a multi- 
ple-social-scientist by acquiring the rudiments of the several relevant 
disciplines. On the other hand, those who wish to remain within the con- 
fines of economic analysis must cut their studies to the limitations of their 
method. Tendencies in both directions are observable among economists 
working in the labor field, and the results of their work will eventually be 
more persuasive than abstract argument about the relative desirability of 
the two approaches. 

The heterogeneity of the field raises difficult problems concerning the 
proper scope of the present essay. I have deliberately concentrated atten- 
tion on problems which can be related to the general body of economic 
theory. Areas of study which cannot be approached via economic theory 
have been treated somewhat more briefly. I want to be very clear that this 
allocation of space is not intended as a judgment of relative importance. 
The political and social consequences of trade unionism are probably 
more important than its strictly economic effects. In stressing economic 
problems, I have been somewhat influenced by the fact that this volume 
is addressed to people with a background of economic analysis, and that 
one of its main purposes is to illustrate the interrelations of various fields 
of economic study. There is the further consideration that application of 
political, psychological, and other non-economic types of analysis to labor 
problems is still in its infancy, and one can do little more at present than 
to indicate some of the more important unresolved issues. 

A different kind of problem is presented by the very great volume of 
writing on labor matters. In order to hold the essay within reasonable 



bounds, it has been necessary to concentrate almost entirely on work done 
in the United States within approximately the past fifteen years. Even 
within this area, no attempt will be made to review all of the published 
material. The literature of trade unionism and collective bargaining alone 
is so extensive that even to prepare a bibliography of it would be a for- 
midable task. Fortunately or unfortunately, the scientifically relevant lit- 
erature is a relatively small portion of the whole. The general policy 
followed here is to omit mention of writings by unionists, management 
officials, political leaders, and popular writers generally. I have also had 
to omit any detailed reference to the periodical literature, though much 
relevant material has appeared in this form. Among books, I have slighted 
specialized studies in favor of more general works, and have given prefer- 
ence to books with some analytical framework over purely descriptive 


I, Wages in the Economy 

Interest in wages has been stimulated recently by the longest and 
sharpest increase of money wage rates in American economic history. 
Average hourly earnings in manufacturing, which stood at $0,437 per 
hour in October 1932, had increased to $1,227 P er hour by June 1947. 
The rise of wages was almost continuous over this fifteen-year period, 
being interrupted only for a short time during 1938. Particularly sharp 
increases in the general wage level occurred during 1933-34, *937> 
1941-42, and 1946-47. The economic consequences of these increases 
were debated almost continuously in both political and academic circles. 

During this period also the wage structure of such basic industries as 
steel, automobiles, electrical products, textiles, rubber, and petroleum 
products began to be determined primarily by collective bargaining. This 
led naturally to active discussion of the probable effects of a bargained 
vvage structure. Another significant development was the marked in- 
crease in the quantity of wage data available. The wage studies of the 
Bureau of Labor Statistics were considerably enlarged, first as a conse- 
quence of the N.I.R.A. and other recovery programs of the 'thirties, and 
again during the wage stabilization program of World War II. In spite of 
recent reductions in B.L.S. activities, we have today an unprecedented 
amount of information about the details of the national wage structure. 1 

Under collective bargaining it does not seem very useful to regard 

1 The material collected by the Bureau of Labor Statistics is the leading source, but by 
no means the only source of wage information. For a fuller discussion of wage data, and 
also for material pertinent to later sections of this chapter, see L. G. Reynolds, "Research 
in Wages," Social Science Research Council Memorandum No. 4, 1947. 


wage rates as market prices. They are administered prices, i.e., they are 
held constant for considerable periods while other things are adjusted to 
the given wage. They may also in a sense be termed "legislated" prices, 
though the legislation is enacted by private agencies with relatively little 
public control. 

Contemplation of a system in which more and more wage rates are 
determined by collective bargaining suggests at least four types of issue. 
First, what are the objectives of unions and employers in wage bargain- 
ing? The objectives of unions are especially important, since the union 
normally appears as the aggressor— at least during periods of stable or 
rising employment. What are the pressures which influence the union's 
initial wage demands and the amount for which it will actually settle in 
a particular case? 

Second, how has the national wage structure evolved in recent years 
under the stress of collective bargaining and of government wage regula- 
tion? What has happened to wage differences between industries, be- 
tween different firms in the same industry, between low-skilled and high- 
skilled occupations, and between different regions of the country? Can 
one say that the wage structure has been distorted as compared with 
some hypothetical competitive pattern of wage rates because of the differ- 
ing strength of unionism in different sectors of the economy? 

Third, how does the economy adapt itself to the kinds of change in 
money wage rates which occur under collective bargaining? How does a 
general change in wage rates throughout the economy affect the general 
level of prices, output, and employment? How do particular industries 
adapt themselves to a rate of wage change greater or less than that occur- 
ring in the system as a whole? The problem of how a wage change con- 
fined to one firm will affect output and employment in that firm is 
perhaps less important, since under collective bargaining one may expect 
most wage movements to be at least industry-wide. This question has 
considerable analytic interest, however, and a satisfactory way of han- 
dling it would be highly desirable. 

Fourth, there is the problem of developing normative rules concerning 
the desirable movement of wages over time under conditions of shifting 
demand and changing technology. Such rules are necessary to guide 
public policy concerning wages, or even to determine whether any public 
policy is required. It may be that the wage movements which occur spon- 
taneously under collective bargaining produce reasonably satisfactory re- 
sults. But we need some way of determining how satisfactory they are 
and, if public regulation seems desirable, of appraising the direction 
which such regulation should take. 


Discussion of recent work on wages in the United States may usefully 
be organized around these four groups of problems, which we may refer 
to briefly as wage determination, wage behavior, economic adaptation to 
wage changes, and criteria of wage adjustment. 

With respect to wage determination, particular importance attaches to 
an understanding of union wage demands and the pressures influencing 
them. Much useful work has already been done on this matter and much 
more is in process. 2 Slichters volume on the economic policies of unions, 
though oriented mainly toward non-wage policies, explores the attitudes 
of unions toward various systems of wage payment, and the problem of 
cost differentials between union and non-union plants. Further, many of 
the non-wage policies which he analyzes have as direct an impact on unit 
labor cost as the wage level itself. Dunlop has constructed a systematic 
statement of the economic issues which a union confronts in developing 
its wage program, and assembled information on the choices which par- 
ticular union groups have made on these issues. Ross has explored the 
way in which unions formulate their wage demands, and has empha- 
sized the extent to which wage settlements are influenced by the union 
leader's necessity of maintaining the union and his leadership of it 
against rival cliques, other unions, and the employer. His political model 
of the union as a wage-setting institution forms an interesting and useful 
contrast to the economic models of Dunlop and others. Lester has ex- 
plored the conditions under which regional or national collective bargain- 
ing may produce wage standardization of one sort or another. A forth- 
coming book by Lindblom attempts a comprehensive analysis of union 
wage objectives and tactics. One should mention also Palmer's incisive 
analysis of the effect of non-union competition on union wage tactics, 
and Kennedy's recent study of union attitudes toward incentive wage 

These studies have perhaps raised more questions than they have an- 
swered. It appears, for example, that wage demands are usually formu- 
lated by the chief officers of the union, with the membership playing a 
permissive or ratifying role. Yet the leaders presumably put forward de- 
mands which they believe will arouse the enthusiasm or at least the 
approval of the rank and file. What are the channels through which 

2 John T. Dunlop, Wage Determination Under Trade Unions (New York, 1944); 
Van Dusen Kennedy, Union Policy and Incentive Wage Methods (New York, 1945); 
Richard A. Lester and Edward A. Robie, Wages under National and Regional Collective 
Bargaining: Experience in Seven Industries, Industrial Relations Section, Princeton Uni- 
versity (Princeton, 1946); Gladys Palmer, Union Tactics and Economic Change (Phila- 
delphia, 1932); Arthur M. Ross, "Trade Unions as Wage-Fixing Institutions," The 
American Economic Review, September 1947, XXXVII, pp. 566-588; Sumner H. 
Slichter, Union Policies and Industrial Management (Washington, 1941). 


membership sentiment is transmitted to the leaders? How do the leaders 
decide that ten cents an hour this year will keep the members in line 
whereas five cents an hour would leave them discontented? How do they 
judge the feasibility of selling non-wage gains to the membership in lieu 
of wage gains? Again, while there is little indication that union presi- 
dents listen to their research directors or make explicit use of economic 
analysis in formulating their wage demands, it is difficult to believe that 
the economic situation of the industry at the time has no effect on their 
deliberations. Is it not possible that, while union leaders make no calcu- 
lation of demand elasticities and so on, the economic situation may influ- 
ence their thinking in other ways— for example, their estimates of how 
large a wage increase the employers will concede without a fight? 

There is a clear tendency for certain "key bargains" to have an influ- 
ence far beyond the workers immediately affected. Such are the bargains 
between the Steelworkers and the United States Steel Corporation, the 
Automobile Workers and General Motors, the Mine Workers and the 
coal operators' associations, the Rubber Workers and the "big four" rub- 
ber companies. The terms of settlement reached in these key bargains 
spread through the imposition of identical demands on other companies 
in the same industry, imitation of these demands by unions in other in- 
dustries, and wage increases by non-union employers in an effort to keep 
pace with the union scale. While some of these focal points— notably 
wage changes announced by United States Steel— existed in pre-union 
days, it seems likely that widespread unionization has increased both the 
speed and the uniformity with which wage changes are transmitted 
throughout the economy. It would be interesting to test this hypothesis 
by a detailed comparison of wage changes during 1946-48 with other 
periods of general wage change during the 'twenties and 'thirties which 
were less influenced by collective bargaining. 

Management opinion concerning the proper level and structure of 
wages, while clearly less important than it was in pre-union days, still has 
a significant influence on the course of events. A few managements have 
a direct hand in shaping the key bargains which become patterns for the 
remainder of the economy; and managements outside the key group may 
be able to deviate a good deal from the general movement. Indeed, use 
of the term "pattern" suggests much more uniformity than actually exists 
in economy-wide wage changes. A statistical study of the 1 946 movement, 
for example, indicates that even in manufacturing the increases varied 
from zero to more than thirty cents per hour, while the average increase 
was considerably below the nominal pattern of eighteen and one-half 
cents. This raggedness of general wage movements must be caused partly 


by the differing resistances of different managements to union wage de- 
mands, and partly by differences in the judgment of non-union firms as 
to how far they dare lag behind the general advance. 

Some work on management wage policies— largely unpublished as yet 
—has been done at Princeton, Yale, Massachusetts Institute of Technol- 
ogy, and a number of other centers. The main limitation of these studies 
is that, for practical reasons, they have been based largely on interviews 
with company officials. Information obtained in this way is bound to con- 
tain a high percentage of rationalization, since the past decisions being 
investigated were arrived at through discussions which cannot readily be 
reconstructed after the event, and were probably affected by circum- 
stances of the moment which were never recorded and have now been 

Moreover, wages are only one of the many things with which the prin- 
cipal officers of a company have to deal. The crucial decisions may relate 
to pricing, sales strategy, production and inventory policy, and invest- 
ment planning; wage questions, particularly where wages form a small 
percentage of production costs, may be settled in a relatively offhand 
manner. An understanding of wage policy, therefore, requires that it be 
regarded as only one (and perhaps a minor) aspect of the general strategy 
of the firm. What is required, in other words, is an inclusive study of the 
economics of the enterprise. A specialized study limited to wage-setting 
will miss so many central problems of the business that it can yield only 
dubious results. The only published study which approaches the neces- 
sary breadth of treatment is the monograph prepared in the late 'thirties 
for the Temporary National Economic Committee. 3 

Most studies of management wage policies relate to general or "across- 
the-board" changes in the plant wage level. It should be remembered, 
however, that management— alone, or through negotiation with the 
union— is continually adjusting the rates paid for particular jobs within 
the plant. Moreover, under incentive systems, earnings may vary over a 
wide range with no change at all in job rates as a result of variations in 
worker effort, changes in production methods, management's strictness or 
laxity in revising piece rates, and so on. Economists have tended to regard 
the detailed determination of job rates and earnings as minutiae, and 
have left discussion of them largely to the industrial engineers. This atti- 
tude is unwarranted; the rules of procedure which the engineers have 
evolved actually pose some very interesting theoretical problems. For ex- 

3 Douglass V. Brown, John T. Dunlop, Edwin M. Martin, Charles A. Myers, and John 
A. Brownell, Industrial Wage Rates, Labor Costs and Price Policies, T.N.E.C. Mono- 
graph 5 (Washington, 1940). 


ample, the principles used by engineers in evaluating the relative worth 
or' various jobs take little if any account of the principles which econo- 
mists from Adam Smith on have proclaimed as the natural or competitive 
determinants of wage differences between occupations. Yet these appar- 
ently arbitrary and non-competitive rules seem to work reasonably well in 
practice. What is the explanation? Is any set of occupational rates feasi- 
ble provided it is consistently applied? Even more interesting questions 
are suggested by the structure of incentive w&ge systems. Pioneer work in 
this area has been done by Dickinson, who has tried to apply the tech- 
niques of the economist, the psychologist, and the industrial engineer to 
the problems of wage rate determination. 4 

The second main area for investigation is the actual behavior of bar- 
gained wage rates over the course of time. Union objectives are one 
thing; accomplishment is another. It is quite conceivable that the various 
unions, despite their efforts to outstrip each other in wage increases, have 
succeeded only in maintaining about the same relative position. It is even 
possible, though not very likely, that the earnings of unorganized work- 
ers in the economy have risen as rapidly as those of organized workers 
during the past fifteen years. It is quite another thing if the effect of col- 
lective bargaining has been to set up widely differing rates of wage 
increase in the various unionized industries, leading presumably to dis- 
tortion of the wage-price structure and misallocation of resources. In this 
connection the statistician is faced with a nice problem: How is one to 
measure or even detect wage distortion in the absence of the norms of 
proper wage relationships which would be provided by a perfectly com- 
petitive labor market? The absence of such norms makes it difficult to be 
at all certain whether collective bargaining is forcing the wage structure 
closer to or farther away from the "competitive pattern." We may quite 
possibly be approximating competitive norms more closely in some re- 
spects (for example, intra-plant occupational differentials) and departing 
more widely from them in other respects (for example, inter-industry 
differences in wage levels). 

In addition to such conceptual difficulties there are serious deficiencies 
in the statistical data with w 7 hich we have to work. The most satisfactory 
data would probably be base rates of pay (and, for incentive jobs, actual 
hourly earnings as well) for certain key occupations, collected once a 
year or so from a sample of identical establishments selected to represent 
the major industries and major industrial areas of the nation. Even this 
information would be far from perfect because of unavoidable variations 

4 Z. C. Dickinson, Compensating Industrial Effort (New York, 1937); idem, Collec- 
tive Wage Determination (New York, 1941). 


from plant to plant in the content of jobs with identical titles, and also 
because of variations in the quality of the workers employed. The unfor- 
tunate fact, however, is that occupational wage data of any sort are very 
rare— so rare that it is difficult to trace even the broad outlines of the 
national wage structure over the course of time. 5 The familiar series of 
average hourly earnings and average weekly earnings in particular indus- 
tries, while useful in forming rough conclusions about long-run trends in 
workers' real income and general welfare, cannot be said to measure the 
price of labor with any precision because they are influenced by so many 
factors other than changes in base rates of pay. 

In spite of these difficulties, one can point to considerable progress in 
studying the behavior of wages. During the past few years there has been 
much useful exploration of just what is meant by "wages" or "the price 
of labor." It has been pointed out that earnings may be computed over 
different periods of time, that unit labor cost to the employer will behave 
differently from earnings, that both earnings and labor costs are influ- 
enced by various types of fringe payment and working rules as well as by 
base rates, and so on. 6 This recognition of a plurality of wage concepts is 
very important for clear analysis, and reflection on it suggests a number 
of further problems. The question arises, for example, whether the most 
common type of published wage statistics— average hourly earnings for a 
plant or an industry— really measures anything which is either statisti- 
cally precise or theoretically interesting. Again, the fact that the wage 
concept which interests workers and is therefore relevant to labor supply 
differs from the wage concept which is important to employers throws 
some doubt on the appropriateness of constructing labor supply and de- 
mand curves on the same diagram, i.e., using identical units on the wage 
axis. 7 Moreover, most of the theoretical literature on wages seems to 

6 The main bodies of data are: (a) the common labor rates published by the National 
Industrial Conference Board since about 1920, and by the Bureau of Labor Statistics 
from 1926 to 1942; (b) occupational wage rates in those industries— building construc- 
tion, printing, and baking— for which local union scales have been compiled annually 
by the B.L.S. over a long period of time, plus a few industries such as cotton textiles 
which have been surveyed by the Bureau at irregular but frequent intervals; (c) the 
occupational rates gathered on a large scale by the B.L.S. for the National War Labor 
Board during the years 1943-45, and the data on occupational rates in major industries 
gathered through the B.L.S. program of industry wage surveys from 1945 to 1947. This 
material for the period 1942-47 begins to approach in completeness what is needed; but 
there is little in earlier years with which one can compare it, and it is likely that budgetary 
limitations will prevent similarly comprehensive studies from being made at all fre- 
quently in the future. 

6 See particularly Susnner H. Slichter, Basic Criteria Used in Wage Determination, 
Chicago Association of Commerce and Industry (Chicago, I947)> Appendix; and John 
T. Dunlop, of. cit., Ch. 2. 

7 It is not meant to suggest that this is not formally permissible. It clearly is, since the 
worker's scale of units (say, weekly take-home pay) is convertible at will into the employ- 


assume that wages are paid entirely on a time basis. What kinds of con- 
ceptual adjustment are necessary to handle wage payments on an output 
or incentive basis: 5 

There have also been several ground-clearing studies of what the na- 
tional wage structure looks like and how it changes over the course of 
time. Douglas' volume on the history of wages in the United States was a 
pioneer effort in this field. 8 More recendy, Lester has made a careful 
analysis of the "north-south" wage differential, the extent to which it can 
be accounted for on productivity grounds, and the changes in its size 
during recent decades. 9 He has also summarized the evidence revealed 
by wartime studies of the Bureau of Labor Statistics concerning the ex- 
tent of wage differences for allegedly comparable work in particular local- 
ities. 10 This material is both more voluminous and more carefully pre- 
pared than any previous information on the subject. 

A large amount of work on the behavior of wages and related variables 
has been published by the National Bureau of Economic Research, and 
more work of this sort is in process. One should mention particularly 
Wolman's work on the history of wages in the United States, Fabricant's 
studies of productivity, Kuznets' national income studies, and a projected 
series of reports on wages and employment changes in particular indus- 
tries (of which two have already appeared). 11 At the Brookings Institu- 
tion, Bell has endeavored to relate the trend of wages in various industries 

er's scale (say, unit direct labor cost). The point is that, since "wages" in the worker's 
sense can change without any change in "wages" to the employer, the supply and de- 
mand curves become quite unstable, and this raises some doubt as to their usefulness. 

8 Paul H. Douglas, Real Wages in the United States (Boston, 1930). 

9 Richard A. Lester, "Trends in Southern Wage Differentials since 1890," The South- 
ern Economic Journal, April 1945, XI, pp. 317-344; idem, "Southern Wage Differen- 
tials: Developments, Analysis, and Implications," ibid., April 1947, XIII, pp. 386-394; 
idem, "Diversity in North-South Wage Differentials and in Wage Rates within the 
South," ibid., January 1946, XII, pp. 238-262; idem, "Effectiveness of Factory Labor: 
South-North Comparisons," The Journal of Political Economy, February 1946, LIV, 
pp. 60-75. 

10 Idem, "Wage Diversity and Its Theoretical Implications," The Review of Economic 
Statistics, August 1946, XXVIII, pp. 152-159. 

11 Leo Wolman, Hours of Work in American Industry, National Bureau of Economic 
Research, Bulletin 71 (New York, 1938); Solomon Fabricant, Employment in Manufac- 
turing, 1899-1939, National Bureau of Economic Research (New York, 1942); idem, 
Labor Savings in American Industry, 1899-1939, National Bureau of Economic Re- 
search, Occasional Paper 23: November 1945 (New York, 1945); Simon Kuznets, 
National Income— A Summary of Findings, National Bureau of Economic Research 
(New York, 1946); idem, National Income— A Summary of Findings, National Bureau 
of Economic Research (New York, 1945); idem, National Product Since 1869, National 
Bureau of Economic Research (New York, 1946); idem, National Product in Wartime, 
National Bureau of Economic Research (New York, 1945); George J. Stigler, Domestic 
Servants in the United States, 1900-1940, National Bureau of Economic Research, Occa- 
sional Paper 24: April 1946 (New York, 1946); idem, Trends in Output and Employ- 
ment (New York, 1947). 


during the 'twenties and thirties to changes in productivity, prices, and 
output in those industries. 12 

Dun] op has done a large amount of work on the behavior of wages and 
on the theoretical significance of this behavior. The subjects which he 
has investigated include the cyclical behavior of money and real wages 
in the United States and the United Kingdom; the timing and amplitude 
of wage changes in various industries during the 1929-37 cycle in this 
country, and their relation to price and employment changes in each in- 
dustry; the behavior of man-hour output and unit labor costs during this 
same period; the definition and measurement of "labor's share" of na- 
tional income; and the cyclical and secular behavior of various types of 
wage differential— occupational differences within an industry, inter-in- 
dustry differences, geographical differences, differences between union 
and non-union rates, and so on— in the United States. 13 These studies 
provide an unusually clear focusing of statistical material on theoretical 
issues, and suggest a large number of hypotheses for future study. 

The problem of how the economy adapts itself to changes in the 
level and structure of money wage rates is really a complex bundle of 
problems which must be separated out for analysis. As a minimum, one 
must distinguish a wage change confined to a single firm, a wage change 
applied uniformly throughout an industry, and a wage change extending 
throughout the economy. Under each of these headings one could set 
up numerous sub-cases, depending on whether the change in question is 
an increase or a decrease, whether it is large or small, whether it is 
expected or unexpected, whether it occurs at a time of rising, falling, 
or stable demand, and so on. If we had adequate methods for analyzing 
these three types of wage change, we should be well on the way toward 
an understanding of actual wage movements such as occurred in the 
spring of 1946 and 1947. For any such movement can be broken down 
analytically into an average rate of wage change for the economy, 
deviations of particular industries from this rate, and deviations of 
individual firms from the average rate of change for their industries. 

The reactions of the individual firm to wage changes are not handled 

12 Spurgeon Bell, Productivity, Wages, and National Income, The Brookings Insti- 
tution (Washington, 1940). 

13 See particularly the following writings: John T. Dunlop, op. cit.; idem, "Trends 
in the Rigidity of English Wage Rates," The Review of Economic Studies, June 1939, 
VI, pp. 189-199; idem, "The Movement of Real and Money Wage Rates," The Eco- 
nomic Journal, September 1938, XLVIII, pp. 413-424; idem, "The Economics of Wage 
Dispute Settlement," Law and Contemporary Problems, Spring 1947, XII, pp. 281-296; 
idem, "Wage-Price Relations at High Level Employment," The American Economic 
Review, May 1947, Proceedings XXXVII, pp. 243-253. 


very satisfactorily by the existing theory of the firm. This is explained 
mainly by the fact that theorists have concentrated on defining equi- 
librium positions under static or quasi-static conditions, and usually also 
under "long-run" assumptions about the mobility and transformability 
of capital. Hicks, Hart, and others have sketched out the beginnings of 
a theory of business planning under conditions of change and uncer- 
tainty. Much more work will be necessary, however, to develop this 
theory to the point of practical usefulness. 14 

There has been very little investigation of the reactions of particular 
firms to changes in wage levels. The T.N. E.G. monograph already re- 
ferred to, Lester's recent investigations, and some unpublished investi- 
gations by the writer in New Haven, are the only studies which come 
to mind. It is interesting that these studies show little trace of the effects 
which existing theories of the firm would lead one to think a wage in- 
crease must have on production methods, investment, and employment. 15 
The explanation may be partly that the field studies have been super- 
ficial and have not extended over a sufficiently long period of time. A 
more fundamental explanation is probably that the theoretical models 
ordinarily used to deduce the effects of wage change are static models, 
whereas actual wage changes occur under dynamic conditions. Wage in- 
creases, for example, normally occur at a time when the firm's revenues are 
also rising, and the problem of how to finance the increase is not so acute 
as it would be under static conditions. If it were possible to investigate 
thoroughly a sufficient number of cases over a sufficiently long period, 
it would doubtless appear that the effects of a wage change are extremely 
variable, depending among other things on the stage of the cycle, the 
size of the change, the ratio of labor to total cost, and the nature of the 
product market. 

The magnitude of the effects would also be expected to increase with 
the period of time taken into account. If one is interested mainly in 
questions of resource allocation, these delayed effects of a wage change 
are undoubtedly of real importance. But for many purposes, notably 

14 For a fuller discussion of this point see my article, "Toward a Short-Run Theory 
of Wages," The American Economic Review, June 1948. 

15 For a discussion of the significance, or lack of significance, of this fact see Richard 
A. Lester, "Shortcomings of Marginal Analysis for Wage-Employment Problems," The 
American Economic Review, March 1946, XXXVI, pp. 63-82; Fritz Machlup, "Marginal 
Analysis and Empirical Research," ibid., September 1946, XXXVI, pp. 519-554; idem, 
"Rejoinder to an Anti-Marginalist," ibid., March 1947, XXXVII, pp. 148-154; Richard 
A. Lester, "Marginalism, Minimum Wages and Labor Markets," ibid., March 1947, 
XXXVII, pp. 135-148; George Stigler, "The Economics of Minimum Wages," ibid., 
June 1946, XXXVI, pp. 358-365; idem, "Professor Lester and the Marginalists," ibid., 
March 1947, XXXVII, pp. 154-157. 


questions of cycle theory, it is the effects within the first six to twelve 
months that really matter. These effects are probably much smaller than 
might be inferred from the existing theory of the firm. 

The defects in the theory of the firm hamper us also in analyzing 
the effects of change occurring throughout an industry; for despite efforts 
to improve on Marshall, present theories of industry adjustment lean 
very heavily on the presumed adjustments of an average or typical firm. 
Further difficulties arise from the fact that under oligopoly the inter- 
action of cost and price changes becomes very complex and it is probably 
impossible to make any mechanical prediction of causal sequences. 1(J 
On the research side, there has been scarcely any attempt to trace out 
the consequences of changes in the wage level of an industry. Statistical 
correlation of wage changes in particular industries with changes in 
prices, productivity, and employment in those industries do not seem 
to yield very significant results. What is probably required is industry 
case-studies in which thorough familiarity with the history and structure 
of the industry would enable one to read meaning into the statistical 
series. 17 

Most important and most elusive is the problem of how an economy- 
wide wage change affects the general level of prices, output, and em- 
ployment. The difficulty of analyzing these effects is part of the larger 
difficulty of securing an agreed framework for cycle analysis— whether 
to use equilibrium or sequence analysis, what assumptions may reason- 
ably be made about reactions within the system, and so on. To discuss 
these matters here would mean retracing the ground covered in Chapter 
2 of the present volume; the reader is referred to that chapter, and 
particularly to the section of it which deals with cost-price relations. 

Most of the work which has been done on wage-employment relations 
in the economy— for example, by Bergson, Douglas, Keynes, Lerner, and 
Slichter 18 — uses an equilibrium approach. I am inclined to think that 
sequence analysis probably affords a more useful approach to this kind 

16 For a few remarks on this problem see my article, "Relations between Wage Rates, 
Costs, and Prices," The American Economic Review, March 1942, Proceedings XXXII, 
pp. 275-289; reprinted in Readings in the Theory of Income Distribution (Philadelphia, 
1946), pp. 294-313. 

17 Waldo E. Fisher, Wage Rates and Working Time in the Bituminous Coal Industry, 
1912-1922 (Philadelphia, 1932). 

18 A. Bergson, "Prices, Wages, and Income Theory," Econometrica, July-October 1942, 
X, pp. 275-289; Paul H. Douglas, "Wage Theory and Wage Policy," International Labor 
Review, March 1939, XXXIX, pp. 319-359; J. M. Keynes, The General Theory of 
Employment, Interest, and Money (London, 1936); A. P. Lerner, "The Relation of 
Wage Policies and Price Policies," The American Economic Review, March 1939, 
Proceedings XXIX, pp. 158-169; Sumner H. Slichter, "Wage-Price Policy and Employ- 
ment/' ihid., May 1946, XXXVI, pp. 304-318; idem, "The Changing Character of 
American Industrial Relations," ibid., March 1939, Proceedings XXIX, pp. 121-137. 


of problem. The only authors (to my knowledge) who have applied 
sequence analysis to the effects of general wage changes are Bissell, 
Johannesen, and Lundberg. 19 Sequence analysis has its own type of ab- 
straction, and it may well be argued that manageable models will be too 
simplified for any practical use, or even that the "period" concept itself 
is hopelessly artificial. Nevertheless, further experiments in this direction 
should be welcomed. 

In the meantime, the consequences of general wage change remain 
among the most controversial and least understood subjects in economics. 
The issue is clearly of great practical importance. In a highly unionized 
economy, any approach to full employment seems likely to generate a 
rate of wage increase which will lead to some increase in the general 
price level. The question then arises of how rapid and protracted a 
price inflation the economy can tolerate as the price of reasonably full 
employment. There are also interesting questions as to the monetary 
prerequisites for a continuing rise of wages and prices, and as to whether 
such a movement can be stable over any extended period or whether it 
must involve frequent relapses. 

The fourth type of problem suggested above was the development 
of norms of desirable wage behavior over time. Stated most broadly, 
the problem is what kind of wage structure and what patterns of change 
in this structure will be most conducive to full employment of economic 
resources and the most efficient allocation of these resources among 
competing uses. The following are only a few of the questions which 
arise under this heading. Over a period of several decades, should the 
general wage level rise at a rate which just absorbs the gains of technical 
change and leaves the general price level unchanged? Or should the 
wage level rise more rapidly or less rapidly than this? Should the 
general wage level respond to cyclical fluctuations in the level of em- 
ployment, and if so, how? What pattern of wage differentials will pro- 
mote the most efficient adjustment of resources to any given pattern 
of demand and technology— for example, is it desirable that wage rates 
for a particular occupation be completely equalized as among firms, in- 
dustries, and geographical areas? Should changes in wage differentials 
be used as a means of transferring labor and other resources from declin- 
ing areas, industries, or occupations into expanding sectors of the 
economy? Should the wage structure be used as a means for achieving 

19 Richard Bissell, "Price and Wage Policies and the Theory of Employment," Econo- 
metrica, July 1940, VIII, pp. 199-239. 

None of johannesen's work is available in English, but a brief summary of his method 
will be found in Lundberg's book. 

E. Lundberg, Studies in the Theory of Economic Expansion (London, 1937)- 


a desirable distribution of personal incomes, or should this problem be 
handled entirely through the tax and outlay structure? 

Answers to these questions will require the best analytical tools which 
can be developed, a great deal of practical judgment, and perhaps the 
addition of political norms of equity in income distribution, the desirable 
balance between security and progress, and so on. It will probably be 
very difficult to get agreement on rules of the game even among econo- 
mists, let alone public and private officials. Yet unless this can be done, 
there is no basis for appraising the desirability of the wage structure 
which is developing under collective bargaining, or for attempting to 
influence private wage bargains in one direction or another. There is 
already a good deal of public intervention in these nominally private 
wage negotiations, and a secular increase in such intervention seems 
almost certain. The problem is to ensure that increased public inter- 
vention means the application of considered principles of wage settle- 
ment rather than the mere transfer of a power struggle from conference 
rooms in Pittsburgh to conference rooms in Washington. One may even 
hope that, if workable standards of public interest can be developed in 
this field, they may in time have a direct influence on union and manage- 
ment behavior. 

II. Labor Supply and Labor Mobility 

The movement of workers from one employer, occupation, or area to 
another has usually been treated as an aspect of wage determination. 
From certain assumptions about worker behavior there are derived supply 
curves of labor which are used to explain the level of wages and employ- 
ment in a firm, industry, or area. Under collective bargaining the use- 
fulness of this approach is very much reduced. It is still true that 
potential mobility determines the lowest wage which an employer can 
pay and retain his labor force. The effective lower limit to wages, how- 
ever, is the rate which the union can be persuaded to accept. This will 
normally be higher, and often very much higher, than the limit set 
by mobility alone. In this case mobility has no direct bearing on the wage 
level. It retains at most some usefulness in explaining management wage 
decisions under non-union conditions. 

The characteristics of labor mobility are very important, however, 
in connection with what may be termed "labor market engineering"— 
the effort to ensure that the detailed articulation of specialized labor 
demands and labor supplies will go on with a minimum of personal 
hardship and economic waste. Included under this heading are such 


programs as decasualization of irregular occupations, reduction of seasonal 
fluctuations of employment, organization of public employment service 
offices, transfer of workers from depressed to prosperous industries or 
areas, transfer of the chronic surplus of farm population into urban 
employments, and the provision of adequate vocational training and 
guidance for young people. A realistic understanding of how workers 
actually choose jobs is necessary in order to estimate how far manipu- 
lation of wage differentials may be effective in redistributing labor sup- 
plies, and what supplementary measures will be most useful. It will be 
worth while, therefore, to discuss briefly a few of the more important 
aspects of labor mobility: movement into and out of the labor force, 
into and out of employment, between geographical areas, between em- 
ployers in the same area, and between occupational levels. 20 


It is not possible here to go into the various possible ways of defining 
employment, unemployment, and the labor force. It should be noted, 
however, that the problem is of practical importance as well as theoretical 
interest; a shift of definitions can easily make a difference of several 
million in the number counted as unemployed at a particular time. 21 
It is worth noting also that efforts at precise measurement of employ- 
ment and unemployment have been intensified considerably since about 
1 930." These measures have dealt with numbers of individuals rather 
than with man-hours or efficiency units, and have tended to assume 
that a single figure could be meaningful for all purposes. Most of them 
have also tended, in varying degrees, to understate the amount of unem- 
ployment at a particular time (especially with respect to the definitions 

' M For a more detailed statement of research problems and hypotheses in the field of 
labor mobility, see Gladys Palmer, "Research Planning Memorandum on Labor Mobil- 
ity," Social Science Research Council Pamphlet No. 2, 1947. 

21 On this whole range of problems see the following writings by Clarence D. Long: 
"The Concept of Unemployment," Quarterly Journal of Economics, November 1942, 
LVII, pp. 1-30; The Labor Force in Wartime America, National Bureau of Economic 
Research, Occasional Paper 14 (New York, 1944); The Size of the Labor Force under 
Changing Incomes and Employment (unpublished manuscript prepared for the Con- 
ference on Research in Income and Wealth, National Bureau of Economic Research, 

22 Louis J. Ducoff and Margaret J. Hagood, Labor Force Definition and Measurement, 
Social Science Research Council (New York, 1947); Aryness Joy, "Meaning of Unem- 
ployment Statistics," Journal of the American Statistical Association, June 1941, XXXVI, 
pp. 167-174; Russell Nixon and Paul Samuelson, "Estimates of Unemployment in the 
U.S.," The Review of Economic Statistics, August 1940, XXII, pp. ioi-iii; Arthur 
Reede, "Adequacy of Employment Statistics," Journal of the American Statistical Associ- 
ation, March 1 94 1, XXXVI, pp. 71-80; W. S. Woytinsky, "Controversial Aspects of 
Unemployment Estimates in the United States," The Review of Economic Statistics, 
February 1941, XXIII, pp. 68-77. 


most relevant to "full employment"), and to understate the amount of 
cyclical variability in employment and the labor force over the course 
of time. 

With respect to the size and composition of the aggregate labor force, 
Long's studies indicate that the proportion of people in each age- and 
sex-group who were in the labor force remained relatively stable in the 
United States between 1890 and 1940. The proportion of the total 
population in the labor force fell very slightly, a somewhat larger per- 
centage of work seekers among adult women being more than offset 
by smaller percentages for boys and girls and for older men. As between 
different cities in the United States at the same time, the studies of both 
Long and Douglas indicate a clear inverse relation between income levels 
and the percentage of the population in the labor force. 23 

There is relatively little evidence on the shape of the (instantaneous) 
supply curve for labor in the country as a whole. The issue is whether 
this curve is positively or negatively elastic with respect to money or 
real wage changes, and whether the curve as a whole shifts in response 
to variations in employment opportunities, and if so, in what direction. 
Wage changes do not seem to have any appreciable short-run effect on 
the proportion of the population seeking employment. There is some 
reason to think that wage increases produce a slight decrease in the 
number of hours which people desire to work, particularly in the case 
of married women with family responsibilities. On the whole, however, 
the supply curve of labor with respect to wages can probably be taken 
as substantially vertical over short periods. 

The question whether labor supply fluctuates sympathetically with 
changes in employment opportunities is probably of greater practical 
importance, but the evidence on it is not at all clear. Woytinsky has 
asserted that deep depression increases the labor force by compelling 
wives and children of unemployed workers to enter the market in larger 
numbers, but the statistical calculations advanced in support of this prop- 
osition have been challenged by other writers. Long concludes that the 
relation is slight and is probably in the opposite direction, i.e., a general 
decline in employment is likely to produce a slight decline in the propor- 
tion of the population in the labor force as some of the unemployed 
abandon the search for w r ork which there seems little hope of finding. 24 

23 Paul H. Douglas, The Theory of Wages (New York, 1934); idem and Erika 
Schoenberg, "Studies in the Supply Curve of Labor: The Relation in 1929 Between 
Average Earnings in American Cities and the Proportions Seeking Employment," The 
Journal of Political Economy, February 1937, XLV, pp. 45-79; Clarence Long, The 
Size of the Labor Force under Changing Incomes and Employment. 

24 D. D. Humphrey, "Alleged 'Additional Workers' in the Measurement of Unemploy 


If one could measure labor supply in the sense of available man-hours 
of standard efficiency, this measure would probably be found to fluctuate 
in the same direction as the level of employment. During a cyclical 
upswing, the number of individuals available for employment probably 
increases, both because some people will accept work who would not 
actively seek it and because employers lower their hiring standards to 
take in workers previously regarded as unemployable. It seems likely 
also that the work week will be lengthened somewhat and perhaps 
approach more closely that which would be freely chosen by employees, 
and that disguised unemployment in its various forms will be reduced. 
Opposite tendencies will be set in motion during a cyclical downswing. 
If this hypothesis is correct, the gap between actual and "full" employ- 
ment may be considerably larger during a depression than is indicated 
by the statistics of "visible" or "superficial" unemployment, and the lower 
the level of employment the greater this discrepancy will be. During an 
upswing, the goal of full employment will recede as it is approached— 
though of course not indefinitely— and the number of additional jobs 
required will be considerably larger than it appeared to be at the bottom 
of the depression. 25 


The characteristics of this type of movement are now rather well 
known as a result of the intensive studies of unemployment in this and 
other countries during the 'thirties. 26 One tends to think of "the unem- 
ployed" as a stable group of individuals occupying the same status year 

ment," The Journal of Political Economy, June 1940, XLVIII, pp. 412-419; Clar- 
ence Long, The Size of the Labor Force tinder Changing Incomes and Employment; 
W. S. Woytinsky, Additional Workers and the Volume of Unemployment in the De- 
pression, Committee on Social Security or the Social Science Research Council (Wash- 
ington, 1940). 

25 In support of this, I would be inclined to argue that the surprisingly large wartime 
increase in the visible labor force— from 53.5 million in December 1940 to 63.2 million 
in December 1944— was due to the fact that the 1940 Census, taken at a time of rela- 
tively low employment, seriously understated what might be termed the "full-employ- 
ment" labor force, and consequently the true volume of unemployment. Long, however, 
who has done more work than anyone else on these matters, is of the opinion that the 
wartime increase was due mainly to special and temporary circumstances— military mobi- 
lization and the Selective Service system— rather than to the increased demand for labor. 
See Clarence Long, The Labor Force in Wartime America, pp. 50-55; and idem, The 
Size of the Labor Force under Changing Incomes and Employment, Sec. 4, pp. 4-5. 

26 Ewan Clague, Walter J. Couper, and E. Wight Bakke, After the Shutdown— (New 
Haven, 1934); Daniel Creamer and Charles W. Coulter, Labor and the Shut-Down 
of the Amoskeag Textile Mills, W.P.A. National Research Project, Report No. L-5 
(November 1939); Edward J. Fitzgerald, Selective Factors in an Expanding Labor 
Market: Lancaster, Fa., W.P.A. National Research Project No. L-4 (June 1939); L. C. 
Marsh, Canadians in and out of Work (Toronto, 1940); Gladys L. Palmer and Con- 
stance Williams, Reemployment of Philadelphia Hosiery Workers after Shut-downs 
«* *933-34> W.P.A. National Research Project in Cooperation with University of 


after year. Actually, during times of high employment, there is rapid 
turnover among the unemployed, and a cross-section analysis at any 
time will show relatively few who have been out of work for more than 
a few months. "The unemployed" resemble the changing occupants of 
a subway train rather than water in a stagnant pool. 

A general decline in employment obviously increases the number 
entering the unemployed group and reduces the chances of leaving it. 
But the really significant thing is that the incidence of both layoffs and 
new hirings is very uneven. Layoffs are most numerous among workers 
in heavy industry, workers with relatively little trade skill, those with 
low seniority, the less employable, the very old, and the very young. 
New hirings are confined largely to workers of high employability, in 
the prime of life, with good work experience, and so on. These pre- 
ferred types of workers continue to turn over in much the usual way, and 
even in deep depression there are still many short-term unemployed. 
In addition, however, there accumulates an increasing number of less 
employable people, who have been unemployed for relatively long 
periods, and whose chances of re-employment diminish with the passage 
of time. 

This hard core of unemployment is not immediately affected by 
economic recovery. The first effect of recovery is that fewer people lose 
jobs, rather than that more are absorbed from the unemployed. The 
next effect is likely to be that new entrants to the market and the short- 
term unemployed are able to find jobs more readily. Recovery must 
be well under way before the longer-term unemployed are called on. 
It must be remembered, too, that the unemployed at the bottom of a 
deep depression are a very atypical group as regards geographical location, 
occupational skill, and industrial attachment, as well as personal traits. 
Unless the configuration of demand which develops during the recovery 
is very similar to that which existed during the previous expansion, or 
unless the long-term unemployed are highly mobile, their reabsorption 
may prove very difficult. It is this which makes long-standing unemploy- 
ment a peculiarly intractable problem, treatment of which requires more 
than the creation of an adequate overall demand for labor. 


This matter has also been studied considerably in recent years, and 
it is possible here to mention only a few of the apparent conclusions 

Pennsylvania, Industrial Research Department, Report No. P-6 (January 1939); H. W. 
Singer, ''The Process of Unemployment in the Depressed Areas," The Review of Eco- 
nomic Studies, June 1939, VI, pp. 177-188; idem, "Regional Labor Markets and the 
Process of Unemployment," ibid., October 1939, VII, pp. 42-58. 


From this work. 27 Geographical movement of workers occurs mainly 
during periods of rising employment; and the main economic stimulus 
to movement seems to be inter-area differences in the number of job 
opportunities available. This may mean either that job openings are 
particularly abundant in the new area or that the outlook is particularly 
unfavorable in the old. Those who leave agriculture for urban employ- 
ments, for example, seem to be mainly young people who have no hope 
of becoming farm operators, or older people forced off the land for one 
reason or another. Those well established in agriculture seem willing 
to remain there even at incomes much below those which they might 
earn in the city. The Oxford Economic Institute's studies of migration 
between certain counties in England found that the greater the differ- 
ence in the unemployment ratios of two counties, the greater, other 
things being equal, was the likelihood of migration between them. 
Moreover, the lag between the occurrence of a discrepancy in unem- 
ployment ratios between two areas and an increase in migration between 
them seemed to be only about six months. 28 

The relation between geographical wage differentials and labor mo- 
bility is difficult to evaluate, since wage differences usually occur along 
with differences in job opportunities and other factors. A few things, 
however, can perhaps be said. First, high wages in an area do not seem 
to have very great attractive power unless accompanied by job openings; 
and while we might expect the two to occur together, this will not always 
be the case. Second, most people who have jobs are sufficiently attached 
to their home communities so that they have little interest in jobs else- 
where, even at considerably higher wages. Interest in opportunities 
elsewhere is usually awakened by the loss of employment at home or 
by some personal or family disturbance. Third, even when the person 
is predisposed toward movement for one reason or another, his move- 
ment is about as likely to follow lines of personal contact 29 as it is to 
follow wage contour lines. 

87 The bibliography of this subject is very large. The following are only a few key 
references: Carter Goodrich et al., Migration and Economic Opportunity (Philadelphia, 
!936); Clark Kerr, "Migration to the Seattle Labor Market Area, 1 940-1 942," Univer- 
sity of Washington Publications in the Social Sciences, August 1942, II, No. 3, pp. 129- 
188; Bureau of the Census, Civilian Migration in the United States: December 1941 
to March 1945, September 1945, Series P-S, No. 5; idem, Internal Migration in the 
United States, 1935-1940, April 1944, Series P-44, No. 10; Francis M. Vreeland and 
Edward J. Fitzgerald, Farm-City Migration and Industry's Labor Reserve, W.P.A. 
National Research Project, Report No. L-7 (August 1939). 

28 H. Makower, J. Marschak, and H. W. Robinson, "Studies in Mobility of Labor: 
Analysis for Great Britain," Oxford Economic Papers, May 1939, No. 2, pp. 70-79; 
September 1940, No. 4, pp. 39-62. 

29 It is well known that international migration is largely a group rather than an indi- 
vidual matter. The first to move to the new land send back for their relatives and friends, 


It should be noted also that geographical mobility is quite selective 
with respect to personal and occupational characteristics. Mobility is 
highest among the young, single, and unattached; it is reduced by age, 
family responsibilities, and home ownership. Among occupational groups, 
professional people are much more mobile than any others, followed by 
executives and other white-collar workers. Skilled manual workers appear 
to be somewhat more mobile than the semi-skilled and unskilled. The 
labor force is not a homogeneous mass, all parts of which are equally 
responsive to wage differences or other economic stimuli. 


This sort of movement has always been of particular interest to 
economists because of its presumed influence on wage determination. 
Yet there has been surprisingly little investigation of workers' decisions 
about taking and leaving jobs. In recent years one can think only of 
Palmer's studies in Philadelphia, Yoder's studies of the Minneapolis-St. 
Paul area, Davidson and Anderson's study of a California community, 
Maclaurin and Myers' study of a New England factory city, and the 
investigation of the New Haven labor market now being carried on at 
Yale. 30 

The results of these investigations cannot be described in any detail 
here, but we may note a few of the hypotheses which seem to be emerging. 
First, only a small percentage of the workers in an area— the unemployed, 
the new entrants, and those strongly dissatisfied with their present jobs 
—can really be regarded as in the labor market at any time. Most workers 
already have jobs with which they are reasonably well satisfied. These 
people do not behave like participants in a market; they are members 
of an organization with which they hope and expect to remain indefi- 
nitely- There is an economic basis for this, since both income and 
security tend to be increasing functions of length of service with a 
particular firm. Probably even more important in binding the worker 
to the plant, however, are his established relations with workers and 

and thus whole groups are transferred from a particular community in the old country 
to a particular community in the new. Although similar studies have not been made with 
respect to internal migration, it would probably be found to follow the same general 

30 P. E. Davidson and H. Dewey Anderson, Occupational Mobility in an American 
Community (Stanford University, California, 1937); Helen Hermann, Ten Years of 
Work Experience of Philadelphia Machinists, W.P.A. National Research Project in Co- 
operation with University of Pennsylvania, Report No. P-5 (September 1938); W. R. 
Maclaurin and C. Myers, The Movement of Factory Labor (Cambridge, Mass., 1943); 
Gladys L. Palmer, "The Mobility of Weavers in Three Textile Centers," Quarterly 
journal of Economics, May 1941, LV, pp. 460-487; L. G. Reynolds and Joseph Shister, 
Job Satisfaction and Labor Mobility (New Haven, 1948). 


supervisors, and a preference for an accustomed routine over the perils 
of novelty and experimentation. 

Second, wages are only one of five or six major factors affecting the 
worker's satisfaction with his job. At least equally important are the 
physical nature of the job, the degree of independence and participation 
in deciding how the job is to be done, equitable or inequitable treatment 
in work assignments and promotions, and regularity and security of 
employment. Moreover, his satisfaction with his wage rate need not be 
based on inter-plant comparisons; rate comparisons with other jobs in 
the same plant, and the adequacy of the worker's income to maintain 
his family at its accustomed standard of living, appear to be the com- 
monest determinants of the worker's satisfaction with his wages. It is 
quite unrealistic to assume that the worker's willingness to leave his 
job depends mainly on the wage level of his plant relative to other plants, 
as is implied in the usual labor supply curves. 

Third, workers who are in the market because of unemployment or 
dissatisfaction with their present jobs seek wcrk primarily through rel- 
atives, acquaintances, former employers, union officials, and other per- 
sonal contacts. The choice of a new job is usually based on a very limited 
comparison of alternative opportunities, and in many cases the worker 
takes the first job he hears about. This is due partly to a widespread 
feeling among workers that jobs are scarce and that one should seize 
the bird in the hand, which apparently persists even during the present 
period of high employment. It is due mainly, however, to the fact that 
the nature of the market makes job shopping very difficult. There is no 
central point at which workers can learn about the full range of job 
openings in the city, employers' use of the local office of the state employ- 
ment service usually being insufficient to accomplish this end. Moreover, 
many of the most important things about a job— treatment by super- 
visors and fellow-workers, the detailed content of the job, the pace of 
work, the actual personnel practices of the company— cannot be learned 
until one has worked on the job. The only way in which a worker can 
really shop the market is to try out one job and then, if necessary, quit 
it and try again. While this process may well lead the worker from 
poorer to better jobs, there is no reason why it should lead him to the 
best job open at a particular time. 

Fourth, the job choices of young people entering the labor market 

seem on the whole to be less systematic and well informed than the 

choices made in later life. There is little calculation of long-run advan- 

tage, and often no comparison even of the immediate advantage of 
alternative jobs. A large proportion of the first jobs turn out to be blind 


alleys, and many youngsters must experiment again and again before 
settling into permanent employment. There is no assurance that this 
process will secure the allocation of labor supplies which would have 
resulted from informed foresight. 

These brief and over-simplified remarks are perhaps sufficient to 
raise some doubts concerning such concepts as the supply curve of labor 
to the firm, the labor market, a competitive wage structure, and so on. 
Economists have tended to regard rational behavior by workers as 
synonymous with complete responsiveness to wage differentials. This 
is too limited a view. Given their preference scales and their conception 
of the alternatives actually open to them, most workers make a rational 
job adjustment. But their conception of their situation is quite different 
from that which might be held by an omniscient observer, and their 
preference systems embrace a great many things besides hourly rates of 
pay. The fact that workers do not jump constantly from job to job in 
pursuit of maximum earnings proves nothing about their rationality, 
though it may prove something about the rationality of economists who 
expect them to do so. 


Under this heading arise two different but related problems. Of major 
long-run importance is the question how freely children of parents in one 
occupational stratum are able to enter other occupational strata on the basis 
of individual aptitude and preference. How firm are the barriers between 
non-competing groups and social classes, and are these barriers tending 
to increase or decrease in importance? Considerable interest attaches 
also to those shorter-range occupational movements which can be achieved 
within the working life of the individual. What are the chances that a 
person launched into a particular occupational level in early life will 
be able to move to some other level? What are the channels of vertical 
mobility and what are the principal obstacles? How closely is vertical 
mobility related to occupational wage differentials, either as cause or con- 

The question of long-range occupational movement has been studied 
but little in recent years. The principal works have been Taussig and 
Joslyn's study of the origins of American business leaders, the study 
by Davidson and Anderson already referred to, and the Lynds' qual- 
itative analysis of social mobility in an Indiana factory city. 31 More 

31 P. E. Davidson and H. Dewey Anderson, Of. cit.; Robert and Helen M. Lynd, 

Middletown (London, 1929); idem, Middletown in Transition (New York, i937)i 
P. A. Sorokin, Social Mobility (New York and London, 1927); F. W. Taussig and C. S. 
Joslyn, American Business Leaders (New York, 1932). 


thorough investigation, which is much to be desired, would probably 
reveal conflicting tendencies at work. One might find, for example, 
that it is becoming easier for the children of manual workers to enter the 
lower grades of white-collared work, and that this is partly responsible 
for the relative decline of clerical salaries relative to wages for manual 
labor. At the same time, entrance to the higher levels of business and pro- 
fessional employment may be growing more difficult as a result of rising 
educational requirements and educational costs. Taussig's study revealed 
a marked and growing tendency for business proprietors and officials to 
be recruited from the children of business families. He was careful to 
point out, however, that this can be ascribed to interclass differences of 
opportunity only on the assumption that differences in inherited charac- 
teristics are not significant— a matter on which there is still no general 


There has also been little investigation of the shorter-range currents of 
occupational movement, 32 and significant work might be done on this 
subject. Taking into account only the manual occupations, it is likely that 
the skilled trades are tending to become a closed group and that the 
chances of movement from semi-skilled to skilled work are declining. The 
field of unskilled and semi-skilled work contains, of course, an enormous 
range of jobs, varying widely in wage levels and in total attractiveness. 
Movement of workers among these jobs is probably affected by a wide 
range of factors, including individual abilities and preferences, training, 
length of service, union rules, personal contacts, and the accident of being 
around when a vacancy occurs. A significant fact is that most specialized 
factory jobs are learned by doing them. New men are trained only as 
vacancies occur, and it is therefore not possible for surpluses of trained 
workers to develop and "beat down" the wage level of the job as is some- 
times assumed in theoretical arguments. 

For the economy as a whole, therefore, the upper limit of occupational 
wage differentials is largely a matter of custom or regulation, though for 
an individual employer it may be a matter of enforced conformity with 
practice in other plants. The situation will be equally stable whether 
machinists are paid 50 per cent more than laborers, 100 per cent more, 
or 200 per cent more; and which of these things will happen is not deter- 
mined by competitive forces. There is doubtless a lower limit which is 
in part competitively determined. 33 A reduction of the differential be- 

32 P. E. Davidson and H. Dewey Anderson, op. cit.; W. S. Woytinsky, Labor in the 
United States (Washington, 1938); idem, Three Aspects of Labor Dynamics (Wash- 
ington, 1942). 

88 For an illuminating discussion of this matter see A. Bergson, The Structure of Soviet 
Wages (Cambridge, Mass., 1944). 


tween laborers and machinists to 10 per cent might make recruitment of 
machinists difficult, at least until such time as people came to regard the 
new situation as normal. 34 It seems probable, however, that occupational 
differentials in the United States are still somewhat wider than is neces- 
sary on grounds of recruitment, for one rarely hears of any difficulty in 
inducing workers to move up from less skilled to more skilled jobs. 

III. Trade Unionism and Labor Relations 

The past fifteen years have seen an unprecedented growth of union 
membership in the United States. The number of union members has 
increased from some three million in 1932 to about fifteen million at 
the present time. About half of the wage and salary earners eligible for 
union membership are now in unions, and it seems likely that this pro- 
portion will increase further in the future. Already collective bargaining 
has a dominant influence on wages and other terms of employment not 
only in the unionized sector of the economy but in the non-union sector 
as well. 

The implications of this development are so important that adequate 
discussion of them would require a separate chapter. It is possible here 
to mention only a few of the key issues in this field. The warning issued 
at the beginning of the chapter should perhaps be repeated at this point: 
brevity of treatment is not meant to indicate that the matters treated are 
unimportant. It represents primarily limitations of space plus the fact that 
on some very important subjects relatively little work has yet been done. 


Studies of unionism fer se may be divided into two broad categories: 
first, efforts to trace the development of a particular union over the course 
of time; second, attempts to generalize about some aspect of unionism 
from the experience of a number of unions. The first type of study has 
thus far been commoner than the second, possibly because a single union 
affords a more clear-cut and manageable area of work, and also because 
such studies are or should be a prerequisite to generalizing studies. 

34 If wages served the function imputed to them by Adam Smith (and most subsequent 
economists) of equalizing the total attractiveness of different jobs, the machinist's wage 
should clearly be less than that of the laborer. The perverse behavior of occupational 
wage differences— i.e., the fact that the jobs which are paid more highly are also more 
attractive in most other ways— is probably to be explained on the ground that wage rates 
do not and are not meant to serve an equalizing purpose. High wages are rather a prestige 
index attached to jobs which have come to be regarded as "superior" through a consen- 
sus of management opinion, influenced somewhat by opinions of workers and (increas- 
ingly) trade union officials. 


The union histories which have been written vary greatly in thorough- 
ness, objectivity, organization, and general quality. Many of them lack 
a clear conceptual framework and thus do not really provide raw material 
for generalizations about unionism as a whole. They remain discrete 
assortments of facts instead of cumulating into a developing body of 
knowledge. Exceptions occur sufficiently often, however, to justify hope 
for the future. One might mention as examples McCabe's volume on the 
pottery industry, Haber's study of the building trades, Seidman's history 
of the needle trades, Hill's study of the teamsters, Lahne's book on the 
cotton mill workers, Jensen's study of the lumber industry, and Loft's 
volume on the printing trades. 35 More such studies, which relate the 
union to the structure of its industry and which attempt to explain his- 
torical developments instead of merely recounting them, are much to be 

Comparative or generalizing studies may be divided broadly into studies 
of trade union government, trade union objectives and policies, and the 
collective bargaining and political tactics of unions. Each of these areas, 
of course, embraces a large number of subordinate problems. Thus the 
first area includes numerous problems which arise in the government of 
a single union— securing efficient administrative organization, maintaining 
a proper balance between the authority of national and local officials, 
ensuring an adequate voice for the membership in the development of 
union policies, protecting individual members against arbitrary expulsion 
or other forms of discipline, and so on. There are also numerous problems 
of inter-union relations, including jurisdictional disputes, the controversy 
over craft vs. industrial organization, and the relation of individual inter- 
national unions to the top federations. 

Among works on trade union structure and government one should 
note particularly the studies by Galenson and Seidman, and the some- 
what older volumes by Saposs and Hardman. 36 A major study of repre- 
sentative government in trade unions is now in process under Leiserson's 
direction, and the results should greatly advance our knowledge on this 
subject. The internal politics and administration of unions have also 

33 David McCabe, National Collective Bargaining in the Pottery Industry (Baltimore, 
1932); W. H. McPherson, Labor Relations in the Automobile Industry (Washington, 
1940); William Haber, Industrial Relations in the Building Industry (Cambridge, Mass., 
I 93°); S. E. Hill, Teamsters and Transportation (Washington, 1942); H. Lahne, The 
Cotton Mill Workers (New York and Toronto, 1944); Joel Seidman, The Needle Trades 
(New York, 1942); Vernon Jensen, Lumber and Labor (New York, 1945); Jacob Loft, 
The Printing Trades (New York, 1944). 

36 Walter Galenson, Rival Unionism (New York, 1940); Joel Seidman, Union Rights 
and Union Duties (New York, 1943); David Saposs, Left-Wing Unionism (New York, 
1926); J. B. Hardman, ed., American Labor Dynamics (New York, 1928). 


been discussed considerably in the periodical literature, and something 
like a political science of union operations is in process of development. 37 
There has not yet been a definitive analysis of the major structural change 
in the American labor movement during the past fifteen years, i.e., the 
great increase in the number and membership of industrial unions both 
within the CIO and the AFL. 38 Most prewar writing on the subject 
tended to highlight the more dramatic incidents in the CIO-AFL cleav- 
age and, in greater or lesser measure, to choose sides in the controversy. 
The time has perhaps come for a more dispassionate treatment which 
would view this controversy as incidental to a major structural shift simi- 
lar to that which occurred much earlier in the British, Swedish, and many 
other foreign labor movements. 

One major category of union objectives, those having to do with wages, 
was discussed in Section A. With respect to non-wage objectives, Slich- 
ter's comprehensive study is the most important to appear in recent 
years. 30 A number of more specialized studies have appeared as doctoral 
dissertations, and the previously cited volume by Millis and Montgomery 
provides an excellent review of the literature on this as well as on other 
aspects of unionism. There is still room for much study of union objec- 
tives and policies, particularly those of the newer industrial unions. 
Casual observation suggests that some types of union rule which have 
been highly valued by many craft groups and which have played a 
prominent role in the literature— for example, limitations on number of 
apprentices and restrictions on technological change— are of much less 
interest to industrial unions. A careful documentation of differences in 
objectives between the older and newer unions, and of the apparent rea- 
sons for these differences, would be very valuable. 

The problem of union objectives in the broader sense of reform of the 
existing industrial order vs. drastic alteration of the existing order, and 
the related problem of union reliance on collective bargaining vs. reliance 
on legislation, have been relatively little studied since the appearance of 
Perlman's Theory of the Labor Movement and the latter volumes of the 

37 Philip Taft, "Understanding Union Administration," Harvard Business Review, 
1946, XXIV, pp. 245-457; idem, "Opposition to Union Officers in Elections," Quarterly 
Journal of Economics, February 1944, XVIII, pp. 246-264; idem, "Democracy in Trade 
Unions," The American Economic Review, May 1946, XXXVI, pp. 359-369; Theresa 
Wolfson, "Union Finances and Elections," Annals of the American Academy of Political 
and Social Science, November 1946, CCXLVIII, pp. 31-36. 

38 See, however, the discussion in H. A. Millis and R. E. Montgomery, Organized 
Labor, Vol. 3 of Economics of Labor (New York and London, 1945); Herbert Harris, 
Labor's Civil War (New York, 1940); and Everett M. Kassolow, "New Patterns of Col 
lective Bargaining," in Insights into Labor Issues, R. Lester and J. Shister, eds. (New 
York, 1947). 

80 Sumner H. Slichter, Union Policies and Industrial Management. 


Commons history. The most suggestive work which has appeared relates 
to European labor movements, though it is not without relevance for 
the United States. 40 In a forthcoming revision of his Theory, Perlman 
will examine the question of whether recent events in the United States 
and elsewhere require any modification of his original thesis. The growth 
of organized labor as an interest group fully comparable in strength to 
industry and agriculture, and the choice which labor makes between 
pursuing its objectives through the agencies of government and pursuing 
some of the same objectives through private collective bargaining, will 
obviously have great influence on the future development of our govern- 
mental institutions. Careful analysis of the political objectives and tactics 
of organized labor in this country, and comparisons with other countries 
in which these tactics have been much more highly developed, provides 
one of the most interesting avenues of labor study. 


Even more intriguing than analyses of unionism fer se are the prob- 
lems arising out of the collective bargaining relationship. Here one en- 
counters socio-psychological questions of conflict between basic attitudes 
of union and management officials, administrative problems of collective 
bargaining procedure, legal and political issues of how the relative rights 
and duties of unions and managements should be defined, and of how 
government may best contribute to the prevention or adjustment of in- 
dustrial conflict. 41 Active study of these matters is under way at a number 
of university centers. Only a few of the more extensive investigations 
now in process can be mentioned here. 42 

Kerr and others at California have in process a series of studies in 
collective bargaining, with particular reference to the patterns of bargain- 
ing which have developed on the Pacific Coast. Harbison and his associ- 
ates at Chicago are exploring a variety of situations in particular com- 
panies in an effort to detect the basic factors influencing the course of 
union-management relations. At Yale, Bakke has investigated the objec- 

40 Adolf Sturmthal, The Tragedy of European Labor, 1918-1939 (New York, 1943). 

41 To anyone with a flair for observing and dealing with human beings under curious 
and ever-changing circumstances, this complex of issues will probably always remain the 
heart of "the labor problem." It is not without significance that many of the ablest labor 
scholars have become deeply involved in these matters via arbitration work and service 
on governmental tribunals. Unfortunately, some of those who have done most in this way 
have written least about their experiences. 

42 A more complete listing of studies under way on this and other labor matters will be 
found in the compilation prepared annually by the Labor Market Research Committee 
of the Social Science Research Council. The most recent report, entitled "University 
Research Programs in the Field of Labor," was issued in February 1947. 


tives and attitudes of union and management officials and analyzed the 
possibility of reconciling these objectives; Chamberlain, in addition to an 
earlier volume on collective bargaining procedures, has studied the issue 
of alleged union encroachment on 'managerial prerogatives" in a number 
of basic industries; and Lindblom has completed a monograph on the 
question of how far bargained wage structures interfere with the adjust- 
ments necessary in a competitive and changing economy. Other studies 
in this general area which deserve mention include Pierson's discussion 
of collective bargaining systems, Tellers quasi-legal analysis of manage- 
ment functions under collective bargaining, the Twentieth Century 
Fund volume on recent collective bargaining developments in particular 
industries, edited by Millis, and Selekman's recent study of union-man- 
agement relations. 43 

Several studies of collective bargaining have been focused primarily 
on "union-management co-operation," though this phrase seems to mean 
somewhat different things to different people and stands in need of 
sharper definition. Nyman's study of the Naumkeag Steam Cotton Com- 
pany, and Smith and Nyman's study of numerous other cotton mills, 
dealt with union policy toward and union participation in revision of 
work loads and incentive rates. 44 Slichter's study of union policies dealt 
at length with union participation in the development of improved pro- 
duction methods which was initiated on several railroad systems, in the 
men's and women's clothing industries, and in a number of other places 
during the nineteen-twenties. 45 The National Planning Association is at 
present investigating some ten or twelve instances of unusually harmoni- 
ous industrial relations in an effort to determine the main factors respon- 
sible for the absence of conflict in these cases, and to discover whether 

43 E. Wight Bakke, Mutual Survival (New Haven, 1946); Neil Chamberlain, Collec- 
tive Bargaining Procedures (Washington, 1944); idem, The Unions' Challenge to Man- 
agement (New York, 1948); idem, "Grievance Proceedings and Collective Bargaining," 
in Insights into Labor Issues; F. H. Harbison, "Some Reflections on a Theory of Labor- 
Management Relations," journal of Political Economy, February 1946, LIV, pp. 1-16; 
idem, "A Plan for Fundamental Research in Labor Relations," The American Economic 
Review, May 1947, XXXII, pp. 375-383; idem, Patterns of Union-Management Rela- 
tions (Chicago, 1947); Clark Kerr, "Collective Bargaining on the Pacific Coast," 
Monthly Labor Review, April 1947, LXIV, pp. 650-674; idem, Robert K. Burns, and 
Robert Dubin, "Toward a Theory of Labor-Management Relations," in Insights into 
Labor Issues; idem and Lloyd Fisher, "Multiple-Employer Bargaining: The San Fran- 
cisco Experience," in ibid.; H. A. Millis, ed., How Collective Bargaining Works (New 
York, 1942); F. C. Pierson, Collective Bargaining Systems (Washington, 1942); Benja- 
min M. Selekman, Labor Relations and Human Relations (New York and London, 
J 947); L. Teller, The Law Governing Labor Disputes and Collective Bargaining (New 
York, 1940); idem, A Labor Policy for America (New York, 1945); idem, Manage- 
ment Functions under Collective Bargaining (New York, 1947). 

44 R. C. Nyman and E. D. Smith, Union-Management Cooperation in the Stretch- 
out (New Haven, 1934); idem, Technology and Labor (New Haven, 1939). 

45 Sumner H. Slichter, Union Policies and Industrial Management. 


these factors are peculiar to the situations in question or whether they 
have transfer value to other situations. 

Much work remains to be done even on the strictly economic conse- 
quences of collective bargaining. 46 Collective bargaining affects the cost- 
price-output structure of industry not only through wage determinations 
but also through rules governing the selection and promotion of workers, 
speed and methods of work, introduction of changes in technology and 
job conditions, and so on. These rules tend in general to impose addi- 
tional costs on the employer in order to protect certain non-wage interests 
of the worker, but the relation between costs and benefits is rather vari- 
able. Some types of rule— for example, grievance procedures and protec- 
tion against arbitrary discharge— may confer great benefits on the worker 
with little cost to the employer or possibly even a reduction in costs. 
Other types of rule may impose considerable costs on the employer with 
relatively little benefit to the worker. It is probably true that under non- 
union conditions the interests of workers were unduly sacrificed to those 
of enterprise owners and consumers. The question is whether collective 
bargaining is currently producing a proper balancing of the interests of 
these groups, or whether the interests of workers are now being unduly 
advanced to the detriment of consumers. 47 

Exploration of this question will probably require case studies of the 
effect of contractual rules and informal union policies on productivity 
and labor costs in particular plants and industries. It is dangerous to con- 
clude too much from the rules themselves. Some of the most important 
practices may not be written into the contract, the contractual terms may 
be administered in ways quite different from what a literal reading of 
them would suggest, and some of them may not be enforced at all in 
practice. Intimate observation of the day-to-day administration of a plant 
is necessary in order to judge the actual impact of unionism on produc- 
tivity and cost levels. 

Discussion of public policy toward collective bargaining centered from 
1937 to 1947 on the operation of the National Labor Relations Act and 
on the recurring proposals for its amendment. Enactment of the Labor- 
Management Relations Act of 1947 opened a new chapter in this discus- 

48 Some of the central issues are presented very succinctly in idem, The Challenge 
of Industrial Relations (Ithaca, 1947); see especially Ch. 2, "The Effect of Trade Unions 
on the Management of Business Enterprises." Slichter's earlier volume referred to above 
is also very relevant in this connection. 

47 The answer to this question will always be a matter of judgment rather than of pre- 
cise measurement. How can one determine precisely how fast workers should work on a 
particular operation? How can one say just how much more consumers should pay for a 
product in order that workers may have pleasanter surroundings in the plant, free recrea- 
tional facilities, rest periods, paid vacations, and other types of benefit? 


sion, which will probably proceed with fresh vigor as the provisions of 
the new law are clarified by administrative action and court decision, and 
as proposals for further legal changes are debated in the political arena. 
The books of Bowman and Rosenfarb 48 provide good summaries of the 
operation of the N.L.R.A. from its passage up to their respective dates 
of publication. Among the more recent discussions of public policy should 
be mentioned the volumes by Slichter and Gregory. 49 

Despite the extensive literature on public policy in labor matters, the 
field bristles with unresolved issues both of general principle and of ad- 
ministrative technique. Three examples only need be cited. First, in what 
respects should the individual worker s freedom of action be protected 
against authority wielded by union officials? Where individual freedom 
conflicts with the requirements for strong and stable unionism, which 
should be given preference? Second, in what sectors of the economy is it 
important that work stoppages be prevented, and what types of govern- 
mental action are likely to be most effective to that end? If one is 
interested in equity as well as peace, the problem of stopping strikes 
turns out to be extremely intricate, and one cannot say that effective 
measures have even been devised, much less given practical application. 
A third kind of issue, raised by Simons, Lindblom, and others, has to do 
with the consequences of industrial peace rather than the prevention of 
industrial strife. Is peace in some cases too dearly bought? Are the collec- 
tive bargaining settlements arrived at by the parties, particularly the wage 
settlements, so disruptive to the economy's operation as to call for some 
public control over their terms? This issue will become increasingly im- 
portant as union organization increases in strength, and might become 
very acute in the event of a vigorous effort to maintain full employment 
through fiscal policy. 

It should be noted that, in addition to federal regulation of labor 
relations, there has been a great extension of state legislation on this 
subject during the past ten years and a very large number of court deci- 
sions in labor cases. There has been no recent effort to integrate all this 
material and to present a unified picture of the current state of labor law. 
The last comprehensive treatise on the subject was published by Witte 
in 1932. 50 Considerable portions of this excellent work have now become 
out of date, and a thoroughgoing review of labor law is very much needed. 

48 Dean O. Bowman, Public Control of Labor Relations (New York, 1942); Joseph 
Rosenfarb, The National Labor Policy and How It Works (New York and London, 

49 Sumner H. Slichter, The Challenge of Industrial Relations; Charles O. Gregory, 
Labor and the Law (New York, 1946). 

60 E. E. Witte, The Government in Labor Disputes (New York and London, 1932.). 



This currently popular phrase is in some ways a very good title for the 
work now to be discussed, since its generality allows it to embrace the 
variety of problems which have interested psychologists, sociologists, 
anthropologists, and others. The problems range over the whole field of 
labor and industrial relations, including such things as the motivation 
of the individual worker (and, for that matter, the business manager and 
the trade union official); the relation between the behavior of these 
people in the plant and outside the plant; the social structure of the 
factory— the formal and informal lines of communication, hierarchy of 
authority, organization of work groups, and so on; worker response to 
various structures of incentives and to various personal and social situa- 
tions, as indicated by output, absenteeism, turnover, and the like; and 
the possible ways of integrating the various groups in the enterprise into 
a more harmonious work team. 

The subject matter, in other words, is identical with that used by 
economists and other students of labor questions. The phrase "human 
relations in industry" connotes, not a separate subject-matter specialty, 
but a different point of view and method of approach. Those adopting 
this approach have in common a distrust of the economist's simple as- 
sumptions about human motivations, as well as his tendency to personify 
such complex entities as the union and the business firm and to reason 
about group behavior on individualistic lines. They are highly conscious 
of the complexity of human motivation, the extent to which the indi- 
vidual's actions are conditioned by his personal history and by the social 
structure within which he acts, and the large area of industrial behavior 
which cannot be explained at all by economic analysis. 

Sharing this critical attitude toward economic methods, they are much 
less agreed on what alternative conceptual systems can usefully be applied 
to industrial situations. No one has yet succeeded in developing a con- 
ceptual framework which fully satisfies his co-workers, 51 and rival sys- 
tems of concepts will probably remain in competition for a long time to 
come. This lack of a settled method of analysis, while somewhat frustrat- 

51 This is explained partly by the fact that the different members of this group have 
started from different disciplinary backgrounds, as is indicated by such varied titles as 
psychologists, social psychologists, industrial psychologists, anthropologists, social anthro- 
pologists, sociologists, and industrial sociologists. It will probably be realized increasingly 
that, regardless of starting points, the problems on which all these people are converging 
are identical problems. Moreover, these central problems of behavior will probably yield 
only to some synthesis of psychological and sociological methods of attack. It is even pos- 
sible that economics may play a modest role in the final synthesis of knowledge on these 


ing to the systematic mind, at the same time provides a challenge to the 
exploration of fresh scientific territory. 

The first substantial impetus to this type of study came from investi- 
gations of the behavior of small groups of factory workers carried on 
co-operatively by the Western Electric Company and a group of research 
workers from the Harvard Business School during the late 'twenties and 
early 'thirties. These studies, which have been exhaustively reported, 52 
suggested that non-economic elements in worker behavior were much 
more important than had been generally realized. During the late 'thir- 
ties and early 'forties work along somewhat similar lines was pursued by 
Chappie and Warner. Chappie attempted to develop a quantitative analy- 
sis of interpersonal relationships in the factory on the basis of frequency, 
regularity, and duration of personal contacts between any two individu- 
als, the person taking the initiative in making the contact, the responsive- 
ness of each person to the other during the conversation, etc., and devel- 
oped some interesting mechanical devices for recording and analyzing 
conversations. 53 Warner, applying an anthropological approach to the 
study of a New England factory city, attempted to describe the broader 
community framework within which the processes of industrial relations 
go on, and to trace the way in which this framework influenced the course 
of strikes and other key events occurring during the period studied. 54 

Investigation along psycho-sociological lines is now proceeding most 
actively at Yale, Chicago, and the Massachusetts Institute of Technology. 
Bakke, who during the 'thirties made pioneer studies of workers' moti- 
vation and the social consequences of unemployment, has since- extended 
his studies to other fields. A study of attitudes toward unionism by work- 
ers, employers, and the public is substantially completed, and there is 
now under way an intensive case study of human relations in a large 
business organization at all levels from the company president to the 
lowest labor classification. Bakke has also developed a systematic theory 
of human behavior for use in the analysis of industrial situations. 55 

Gardner, Whyte, and others at Chicago have, either completed or in 
process, a variety of in-plant studies on such subjects as systematic output 

68 Elton Mayo, The Human Problems of an Industrial Civilization (New York, 1933); 
F. J. Roethlisberger, Management and Morale (Cambridge, Mass., 1946); idem and 
W. J. Dixon, Management and the Worker (Cambridge, Mass., 194O; T. N. White- 
head, Leadership in a Free Society (Cambridge, Mass., 1936). 

53 Eliot D. Chappie and Conrad Arensberg, Measuring Human Relations (Province- 
town, 1940). 

M See his Yankee City series, Vols. 1-4 (New Haven, 1941-47). 

60 E. Wight Bakke, Principles of Adaptive Human Behavior (New Haven, 1946); 
idem, Why Workers Join Unions (New York, 1945); idem, The Unemployed Worker 
(New Haven, 1940); idem, The Unemployed Man (London, 1933); idem, Citizens 
Without Work (New Haven, 1940). 


restriction among factory workers, the relation between variations in 
individual output under an incentive wage system and various social 
characteristics of the workers, the characteristics of the workers joining 
the union in a newly organized plant as compared with those who do 
not join, problems in the relationship of workers and first-line supervisors, 
and problems of customer-worker relationships in stores and restaurants. 
Gardner's recent book provides perhaps the best general exposition of 
the human relations point of view to be found in any single source. 56 
McGregor and others at the Massachusetts Institute of Technology use 
a clinical approach to industrial relations situations in particular plants, 
and are working also on a general theory of behavior into which particu- 
lar observations can be fitted. There are no doubt numerous other studies 
of this type which might be mentioned. Published work is somewhat 
difficult to locate because it may appear in any of a large number of 
social science journals. 

It may be desirable in conclusion to emphasize again one or two of the 
points made in earlier sections. The study of the phenomena of indus- 
trial employment calls for a wide variety of analytic techniques. Few 
problems can be handled primarily by economic analysis, and in most 
parts of the field the role of economics is quite limited. Where a problem 
is amenable to economic methods, however, there is no reason for not 
using the best theoretic tools available, while at the same time trying to 
refashion these tools into more realistic and useful forms. This seems to 
be increasingly recognized, and among the younger men entering the 
labor field the gap between theorists and research workers appears to be 
narrowing. It is probably more true today than it was a generation ago 
that the people most highly trained in deductive analysis are among 
those best informed on factual questions and most active in empirical 

The problems not susceptible to economic analysis are being studied 
increasingly by the use of other social science disciplines, and a serious 
effort is being made to fashion new analytic systems suited to these prob- 
lems. Most of the work on these matters is still necessarily on the pre- 
scientific level of careful observation and description. The outlines of 
the phenomena have to be sketched in before we can say what are the 
key questions to be answered. There is good reason to hope, however, 
that description will be followed before too long by analysis and general- 

GG Burleigh B. Gardner, Human Relations in Industry (Chicago, 1945); William F. 
Whyte, ed., Industry and Society (New York and London, 1946). 



Carl S. Snoup 

The present time happens to be particularly appropriate for review- 
ing the developments of a decade in the field of national income. Ten 
years ago the first of the "Studies in Income and Wealth" was published, 
containing the papers and discussions at the initial sessions of the Con- 
ference on Research in National Income and Wealth, held in 1936. It 
was in 1937 also that there appeared the first of the series of volumes 
on national income by Simon Kuznets under the auspices of the National 
Bureau of Economic Research. A debate was under way in the scholarly 
journals over the implications of defining saving and investment, as 
Keynes had done the year before in his General Theory, so that they are 
equal in amount. 

Ten years later, Kuznets had completed his work with the publication 
of a summary volume, late in 1946. The Department of Commerce, in 
July 1947, had issued its revised series of estimates of national income 
for the period 1929-46. The new series reflects some major changes in 
concepts as well as improvements in the data. In Great Britain the annual 
White Paper on national income had become an established institution, 
the seventh one appearing in 1947. In 1946 a mimeographed report (to 
be available soon as a United Nations publication) was submitted by the 
Sub-Committee on National Income Statistics of the League of Nations' 
Committee of Statistical Experts. The report presented a 100-page memo- 
randum by Richard Stone, "Definition and Measurement of the National 
Income and Related Totals." In 1947 the United Nations had under way 
a project for reconciling inter-country differences of concept, to enhance 
the comparability of the national income series of the several nations. 
In the autumn of that year, workers in the national income field from 
many countries attended the International Statistical Association and 
Econometric Society meetings, and organized an International Associa- 
tion for Research in Income and Wealth. By 1947, a few courses in 
national income were being offered in American universities, and exten- 
sive use was being made of J. R. Hicks and Albert Gailord Hart's volume, 



The Social Framework of the American Economy, an adaptation of a 
similar volume by Hicks published earlier in England. A general treatise 
on national income by the present writer, Principles of National Income 
Analysis, designed in part to meet the needs of students specializing in 
national income, was published in 1947. Although the coming decade 
will probably show an increase, not a slackening, in the amount of time 
and effort spent on national income analysis in almost every country, 
including the United States, it cannot quite duplicate the excitement 
and the hazards of the pioneering period just concluded. 

I. The Organizations Concerned 

As the experience of this decade has shown, the study of national 
income has reached a stage where substantial advances in assembly and 
interpretation of the data usually require resources on a scale so large 
that some organization— foundation, research bureau, or government— 
must participate. The same is true to a considerable degree even in the 
refinement of conceptual issues; criticism, rejoinder, and further explora- 
tion have flourished through the medium of conferences, held at more 
or less regular intervals, that bring together the geographically scattered 
workers in this field. Consequently, a resume of developments during 
the past ten years may start with an account of organizational activity 
in national income research. 

The Income Conference, first held in 1936, as noted above, was origi- 
nated and has been maintained by the National Bureau of Economic 
Research. Membership in the Conference has been by invitation, ex- 
tended to persons who have been actively interested in national income 
research. Each member is entitled to vote at meetings and elections of 
the Conference, to receive reports and publications of the Conference 
(subject in some cases to charges), and to be reimbursed for his expenses 
in traveling to and from and in attending the Conference, if arrange- 
ments cannot be made to have these outlays met by the agency or organ- 
ization with which he is connected. In 1947, owing to the growing 
number of members, it was decided that only the expenses of actively par- 
ticipating members (authors and discussants) could be thus defrayed, in 
the future. Membership has not been static; new members have been 
added, and invitations have been discontinued to those whose interest has 
shifted away from national income research. The annual conferences 
have also been attended by non-members upon invitation. 

Although this restriction of the annual conferences to those receiving 
invitations gives rise to certain dangers of exclusiveness and inbreeding, 


it has probably been an essential element for the degree of success that 
has been achieved. Progress in a highly technical field of research, in its 
formative period, could easily be hampered rather than helped by 
time- and energy-consuming conferences where the technician and non- 
technician spent most of the day trying to understand one another. Even 
the selected group of research workers engaged in the same field found 
it difficult enough to understand each other when the Conference was 
initiated. The development of a common technical vocabulary has been 
one of the major achievements of the Conference. 

The other phase of the National Bureau's support of national income 
study has been reflected in the series of volumes by Simon Kuznets and 
his associates at the Bureau. In these volumes the national income data 
for 1919-38 have been assembled and interpreted (some data are pre- 
sented for the period since 1869). 

The Federal Government entered the field of national income research 
in the early i93o , s as a result of a Senate resolution directing the Depart- 
ment of Commerce to prepare estimates of national income. A large 
amount of data and interpretative material has been published: in special 
reports and bulletins in the 1930^, and in articles in the Survey of Cur- 
rent Business in the i94o's. The 54-page Supplement to the July 1947 
Survey represents more than five years of intensive work by Milton Gil- 
bert, director of the National Income Division of the Bureau of Foreign 
and Domestic Commerce, and his associates, particularly Edward F. Deni- 
son, George Jaszi, and Charles F. Schwartz. 

The National Industrial Conference Board and the Brookings Institu- 
tion have also carried on research in national income, though not on the 
extensive and continuing scale of the National Bureau and the Depart- 
ment of Commerce. 

In Great Britain it has been a government department, the Central 
Statistical Office, that has played the chief role in developing national 
income data during the past decade, under the guidance of Richard 
Stone. The data, together with a rather limited amount of conceptual 
explanation, have appeared in the White Papers noted above; the con- 
ceptual issues have been discussed at greater length by Stone and others 
in articles in the scholarly quarterlies, especially the Economic Journal. 
In the past few years two privately supported research groups have em- 
barked on work in the national income field, also under the supervision 
of Stone: the National Institute of Economic and Social Research in 
London, and the Department of Applied Economics at Cambridge. But 
so far there have been no regular series of conferences, in Great Britain 
or elsewhere, like those in the United States. 


In the Netherlands the Central Bureau of Statistics has been active 
in national income computations and analysis. Government departments 
in Australia, Canada, and Eire have recently published estimates of na- 
tional income for their respective countries. The Scandinavian countries 
are represented by the two-volume work of Lindahl, Dahlgren, and 
Kock, on the national income of Sweden, published in 1937, for the 
period 1 861-1930, and by a recently published official series for Den- 
mark. A substantial amount of activity is under way in Norway. In other 
countries, too, the collection of national income data has already started, 
but the work is apparently not yet very far advanced. Generally speaking, 
all of this activity is taking place under government auspices. 

II. Advances in National Income Theory in the Past Decade 

National income theory, in the sense understood here, is concerned 
primarily with conceptual problems, as distinguished from a study of the 
measurement and explanation of the reasons for business fluctuations, 
the implications of the existing distribution of income, and similar topics 
that are referred to below under the heading of "Uses Made of National 
Income Data." There is first of all the basic problem of specifying what 
it is that is being measured when a total of national income is cast up, so 
that changes in the total from one period to another, or one place to 
another, can be interpreted correctly. Different totals may of course be 
constructed, each measuring a different thing; but the thing that each 
total measures needs to be described clearly, and that is not so simple a 
task as might at first appear. Moreover, to avoid misunderstanding, it is 
desirable either to attach the term "national income" to but one of these 
totals, devising other terms for the others, or to specify what species of 
national income is being reckoned: for example, national income at fac- 
tor cost, national income at market prices, gross national income. 

Once the basic definitions have been constructed, at least provisionally, 
there must be examined a large number of sub-problems relating to the 
inclusion or exclusion of this or that item, in order to avoid double count- 
ing, or to avoid gaps, or to assure consistency with the deflating technique 
(product-price index) that is to be applied to the year-by-year or nation- 
by-nation series of totals, or to maximize the usefulness of the series for 
economic policy decisions. 


Great progress has been made in the past ten years in developing uni- 
form and distinctive labels for various totals, especially through the 


efforts of the government research staffs in the United States and the 
British Commonwealth. The progress in making quite clear all the impli- 
cations of each of the totals has been somewhat less satisfying. Much of 
the advance in this area has come in the course of discussing issues raised 
by the various constituent items— a natural enough procedure, but one 
which, as indicated further below, has left the analysis of the total con- 
cepts in a somewhat more fragmentary and disorganized state than might 
have been hoped for. And it may be argued that progress on many of the 
sub-problems raised by the constituent items would have been more rapid 
if, concurrently, more attention had been paid to what the changes in the 
total or totals were supposed to measure: for example, changes in total 
output, or total input; changes in welfare (however defined), or changes 
in economic power. Similarly, much of the discussion of whether part or 
all or none of government services should be included in national income 
would have been more fruitful if the discussants had first reached a more 
precise agreement, or agreement to disagree, on the attributes of a final 
product (as contrasted with an intermediate product), since national in- 
come is the sum of the final products turned out during the period. Like- 
wise, some of the argument over the treatment of taxes in computing fac- 
tor payments would have produced more light than it did, if it had been 
accompanied or preceded by more thought on the type of price index 
needed to make periods and countries comparable. This point of view 
may not be shared, however, by many of those who have made the most 
outstanding contributions to the solution of the issues raised by the vari- 
ous constituent items. Instead, it appears that in government research 
circles, particularly, whether in the United States or abroad, attempts at 
precise description and analysis of what each of the totals is supposed to 
measure are considered as rather less urgent tasks than the careful and 
orderly presentation of constituent elements in a way that allows the 
reader to construct any one of several totals himself, and, especially, 
allows him to use the relationships between the constituent items in the 
study of problems of current economic policy. Thus the League of 
Nations' sub-committee points out that "where national income studies 
are used in connection with the formulation of economic policy, . . .it 
is the interrelationship of transactions that is important rather than indi- 
vidual totals, such as the national income, or gross national product." The 
totals are obtained by "a suitable combination of these constituent trans- 
actions." 1 Richard Stone, in his accompanying memorandum, says that 

1 League of Nations, Report of the Sub-Committee on National Income Statistics, 
Measurement of National Income and the Construction of Social Accounts, mimeo- 
graphed, p. i. 


''modern inquiries which had their origin in an attempt to measure cer- 
tain broad totals have changed their emphasis and now concentrate more 
on the structure of the constituent transactions and on the mutual inter- 
dependence of these transactions." 2 The question at issue, consequently, 
is whether the formulation of standards for the "suitable combinations" 
has been given adequate attention. 

The discussions on the meaning of the totals that have marked the 
past decade resist ready summarization. Several debates on this issue can 
be found in the proceedings of the Conference on Income and Wealth, 
and Kuznets has devoted chapters to it in his volumes on national income 
and national product, but there is still room for a more thoroughgoing, 
unified analysis. It is not even clear, in some discussions, whether changes 
in the totals are to be taken as measuring changes in output or changes 
in input; if the latter can be measured, it presumably requires the use 
of an index of factor prices, not product prices. And if a change in out- 
put is being measured as an indication of change in consumers' welfare, 
the theoretical requirements for being able to ascertain whether such a 
change has occurred need to be developed more than they have been 
in the American literature. On this point, fortunately, the British ana- 
lysts have given substantial aid, especially J. R. Hicks in his article in 
Economica for May 1940. There he distinguished between a welfare 
total and a productivity total. Both of them are output, not input, con- 
cepts, but they differ in that the welfare total does include, and the pro- 
ductivity total does not include, indirect taxes and other payments that, 
not being considered a part of factor payments, simply drive a wedge be- 
tween the market value of the product and the total of factor payments. 
These two measures can, and ordinarily will, give different quantitative 
results when the two periods or two countries that are being compared 
with respect to total national income show movements in product prices 
and in amounts consumed that are not the same, product for product, in 
the two periods or countries. More analysis would seem to be needed, 
however, before it becomes indisputably clear what it is that the pro- 
ductivity index measures. 

As the remarks above imply, the interpretation of the total for national 
income is in large part a problem in the construction of index numbers. 
The total, standing by itself, means nothing; it is compared with its con- 
stituent parts, or it is expressed on a per-person basis and compared with 
prevailing prices, or it is compared with totals for other periods or other 
regions. In this last case, deflation by an index number of product prices 
is essential. The theoretical groundwork for the construction of such an 

3 Ibid., Appendix, p. 23. 


index has been advanced in the past decade by explorations into the 
theory of consumer choice and consumer welfare, particularly in Great 
Britain. This theory, as presented by Kaldor and Hicks, shows that an 
unambiguous answer as to whether the consumer is better off or worse 
off, after prices and his money income have changed, requires that, in 
addition to an assumption that his wants have remained unchanged, 
there be changes in prices and money income of a nature that allows 
him to make a choice between the two assortments of goods— the assort- 
ment that he purchases in the first period and the assortment that he 
purchases in the second. This choice must of course be present only in 
one of the two periods (otherwise he would not in fact buy different 
assortments in the two periods). The assortment that he prefers is the 
one he buys when he has the chance to buy either. This reasoning is 
then extended to a group of consumers, without making interpersonal 
comparisons of utility, by ascertaining whether it would be possible, by 
a redistribution of the goods, to leave each person better off than in the 
earlier period; if so, national income in the sense of consumer goods and 
services is said to have increased. 

In the United States, the Department of Commerce has well under 
way the construction of index numbers for deflating various parts of its 
product series. Kuznets, in his National Product Since 1869, has made 
use of price indexes that are more suitable because more comprehensive 
than those that were available when his earlier estimates were made. 
Finally, the puzzling problems of deflation raised by the shift from peace- 
time goods to wartime goods after 1940 provoked a lively discussion in 
the American field which, while it hardly settled all the issues (and 
some of them seem capable of resisting settlement indefinitely), cleared 
some ground. 

One aspect of attempting to ascertain just what the totals imply is 
seen in the computation of the total for a period shorter than a year. Such 
a computation as that made by Barger, and published in 1942, serves to 
sharpen some of the conceptual problems; so, too, presumably, would a 
computation using a period longer than a year as the time unit. 

Perhaps the most significant change that has occurred during the past 
decade with respect to the concept of a total has been the growing use 
of several types of totals: national income, gross national product, net 
national product, for example. Not all of them were primarily inventions 
of the past decade; the gross national product concept was developed 
before then, but even the gross national product, as it is defined today, 
has no exact counterpart in the earlier literature. 

The two most widely used total concepts are: national income, com- 


puted as the sum of factor payments, themselves computed after the sub- 
traction of indirect taxes but before subtracting direct or personal taxes, 
which are deemed by Kuznets, economists in the Department of Com- 
merce, and the British White Paper to be paid out of the factor pay- 
ments; and gross national product, computed as the sum of net foreign 
investment, net change in inventory, goods and services flowing to con- 
sumers (consumer outlay), goods and services utilized by government 
(government purchases), construction, and production of other fixed 
tangible capital assets for business. The "gross" element arises from the 
fact that the last two items are computed without subtracting an amount 
representing the depreciation or obsolescence of buildings, machinery, 
and equipment. Moreover, the gross national product (except as defined 
by Kuznets), unlike the national income total, nowhere subtracts indirect 
taxes. Since the market value of the finished products listed immediately 
above includes the amounts that flow to government as indirect taxes, 
the gross national product includes an amount equal to such taxes. This 
inclusion, however, is best regarded not as a "gross" element but as a 
method of stating the value of the finished product at a higher price 
level (market prices) than that used in stating the sum of factor pay- 
ments. Opinion on the degree of usefulness of the gross national product 
concept is not unanimous; for example, whether it is a better index of 
activity than is national income. In any event, the extended discussions 
of the gross national product concept over the past decade have helped 
provoke a realization of the issues involved in attempts to "gross up" 
the national income total. 

An intermediate total that may well prove more useful for certain 
purposes than either gross national product or national income is net 
national product, as defined by the Department of Commerce. This, with 
minor qualifications, is equal to national income plus indirect taxes; it 
is the sum of final products expressed at market prices (hence including 
indirect taxes) but with construction and plant and equipment data on 
a net basis, that is, after depreciation and whatever obsolescence the data 

Still a fourth total has recently been devised by the British White 
Paper: gross national product at factor cost, which is, in effect, gross 
national product minus indirect taxes, or, alternatively, national income 
as the sum of factor payments, before deducting depreciation and obso- 
lescence in computing the factor payment, profits. 

The problem of which total to use is especially important when it is 
desired to express each of the components on the product side as a per 
cent of the total. If each of these components is to be expressed as a per 


cent of national income, which does not include indirect taxes, each one 
must be purified by subtracting the amount of indirect-tax loading that 
goes into the market price of each. This is, of course, a formidable task, 
which has so far been essayed only by the British computers. This diffi- 
culty emphasizes one reason for using a net national product concept 
rather than national income, as those terms are defined above. 

Among the totals, that for personal income has been developed and 
refined during the past decade to a considerable extent. This series, for- 
merly known as income payments to individuals (in Department of Com- 
merce terminology), has been expanded in the revision published in 
1947 to include net imputed rent of owner-occupied dwellings and in- 
come in kind paid to the armed forces (value of food, etc., supplied to the 
armed forces). It also now includes an adjustment to remove, from the 
data showing change in inventories held by unincorporated concerns, 
that part of the change reflecting change in price levels. These alterations 
in the personal income series exemplify more the increase in available 
data than an alteration in standard concept. In general, there has been 
perhaps less discussion of what the concept, personal income, should in- 
clude than might have been expected. 


It was indicated above that more substantial advances in national in- 
come theory had been made during the past ten years with respect to 
particular constituent items than in interpreting the totals themselves. 
The greatest advance of all has been in segregating the major sectors of 
the economy that differ notably in the kind of accounts they keep and 
the functions they perform, and, by a system of double-entry bookkeep- 
ing, connecting the sectors by an internally consistent system of accounts. 

With respect to individual items where considerable advances have 
been made during the past ten years, those noted in the following sum- 
mary must be taken as samples rather than a complete listing, which is 
impracticable within the scope of this essay. 

The Government Sector 

The treatment that should be accorded government items in the na- 
tional income accounts has been discussed repeatedly during the past 
decade. Whether substantial progress has been made in this part of the 
field as a result of discussions is pretty much a matter of opinion, for, 
despite the amount of time and effort spent in exploring the conceptual 
aspects of this problem, there is still no general agreement among national 
income analysts. The problem falls into two subdivisions: the treatment 


of government outlays when computing national income as the sum of 
final products, and the treatment of taxes and subsidies when computing 
national income as the sum of factor payments. 

Under the product approach, three schools of thought can be distin- 
guished. The Department of Commerce and the Central Statistical 
Office, along with government statisticians in other countries, consider 
that all government outlays are outlays for final products, not intermedi- 
ate products, and hence should all be counted in the total. The only 
exceptions are outlays that are transfer payments, that is, payments not 
for current goods or services, and hence not for either final or intermedi- 
ate products. Interest on the public debt, or at least the central govern- 
ment debt, is so considered. But police services, fire-fighting services, edu- 
cation, and national defense or national aggression, to name a few of the 
major items, are all considered to be final products. 

The second school of thought, upheld by a few individual students 
in the field, including the present writer, and represented also in the pre- 
war German and the Swedish computations of national income, maintains 
that government services must be divided into two parts: one part con^ 
sists of final goods, either as consumer goods or as additions to the coun* 
try's stock of capital equipment, while the other part consists of inter- 
mediate goods which are used up, or utilized, by government itself or by 
private business, in turning out final products. The distinction is based 
on the particular definition of final product which any one proponent of 
this method has presumably developed, explicitly or implicitly. 

The third point of view is represented by the practice adopted by the 
National Bureau of Economic Research in the early logo's and carried 
on by Kuznets. It assumes that the nature of the government's product 
is best determined, or at least determined with the minimum amount of 
arbitrariness, by noting how much is paid in personal taxes, and how 
much in business taxes. An amount of government product equal to the 
amount of personal taxes paid is considered to be consumer goods. On 
the product side of the national income accounts this item appears, not as 
government product, but as a part of the total consumers' outlay, or flow 
of goods and services to consumers. An amount of government product 
equal to the amount of business taxes paid is assumed to be inter- 
mediate product, hence not included in the national income. Govern- 
ment construction is considered to be a final product. If the total of gov 
eminent outlay is greater than the sum of these three amounts, that is, 
if there is a deficit on current account, this excess, or deficit, is presumed 
to represent expenditures that resulted in no product at all. 

Some reconciliation between these strongly divergent points of view 


may be expected, or at least hoped for, but only if more attention is de- 
voted to what is meant by "final product" in general, and if the propo- 
nents of the three methods specify more clearly than they have so far how 
much of their respective prescriptions reflect conceptual differences, and 
how much they reflect differing estimates of the difficulty of separating 
government outlays into the two classes of final and intermediate product 
on the basis of the data available. 

The other subdivision of the government-sector problem, the treatment 
of taxes and subsidies, has likewise been discussed at length during the 
past decade, and here too disagreement has persisted. But it is perhaps 
less clear than with respect to the product approach how far the disagree- 
ment is real, and how much of it would disappear if the parties to the 
controversy tried first to reach agreement on how the total of national 
income, computed from the factor-payment side, should be defined and 
adjusted by index numbers for year-to-year and place-to-place compari- 
sons. Both Kuznets and the Commerce-White Paper computations sub- 
tract business taxes ("indirect taxes" in White Paper terminology) but 
not personal taxes. Others, including the present writer, are disposed to 
make no distinction between business taxes and personal taxes in comput- 
ing the national income total on the grounds that use of product-price in- 
dexes to make the totals comparable from year to year or place to place 
will adjust for any change in prices caused by a shift from one type of tax 
to another. To this the other group replies that a comparison of the com- 
ponent parts of the national income for any one year— wages compared 
with profits, for example— cannot be made unless a decision is first reached 
as to what taxes come out of factor payments and what taxes are simply 
something added to, but not a part of, factor payments. And to this in 
turn the reply may be that such a comparison is one of relative inputs, 
not output, and so of course requires different rules, or additional infor- 
mation. This highly condensed version of the disagreement is intended 
merely to suggest, rather than define sharply, the nature of the dispute. 
Here again, as with the product side of the government sector, it seems 
likely that a much better understanding would result if some stipulations 
were first made on the broader conceptual issues : whether it is changes in 
(or relative amounts of) input or output that are being measured; if out- 
put, what concept of final product is being employed; and what type of 
price index is to be used in ascertaining the real, as opposed to money, 
changes or relative amounts. There is a widespread reluctance to open up 
once more the discussion on how to treat the government items in national 
income computations; the subject seems to have been worn threadbare by 
repeated papers, discussions, and comments. But it is not very satisfac- 


tory to leave the problems suspended in their present state of semi-solu- 

Saving and Investment 

If the national income analysis and estimates prior to 1937 did not 
explicitly assume that the total of saving could differ from the total of 
investment, neither did they face the issue explicitly and draw, for the 
reader's benefit, the implications that would follow. Keynes' decision, in 
his General Theory, to define saving and investment so that they are 
equal in amount for any past period, forced the pace of analysis in this 
part of the national income field. The structure of the British White 
Papers on national income has obviously been built largely around this 
saving-investment concept. The Department of Commerce has made in- 
creasing use of the identity, and in its revised data published in 1947 it 
presents for the first time an annual series from 1 929 showing total gross 
private saving, government surplus or deficit (saving or dissaving), and 
total gross investment, the sum of the first two items equaling the third 
item. In Kuznets' series, too, total saving equals total investment, al- 
though less emphasis is placed upon this equality— or perhaps it should 
be said that the equality is more taken for granted— than in the other 
two series. For national income analysis, the chief significance of utilize 
ing the Keynesian definitions is probably the stimulus given to "model- 
building" in estimating inflationary and deflationary pressures, or in test- 
ing business forecasts for internal consistency (these points are noted in 
Section IV below). And one of the indisputable benefits is that no one 
can now talk loosely of saving and investment in national income data in 
a way that implies, even aside from statistical discrepancies, the one may 
differ from the other, without being suspected of economic illiteracy 
(excepting, of course, the cases where the analyst makes explicit use of 
the Robertsonian definitions, or some other particular set of definitions 
that he is able and willing to specify). A great deal of potential misunder- 
standing has thus been avoided. 

Inventory Valuations 

The usual method of valuing inventories on the books of business firms 
has been to count them at cost, or, if the market cost of replacing them 
is lower, at that lower value. When prices in general rise, after the firm 
has accumulated an inventory at lower price levels, the firm's sales at the 
new high price level are likely to show a substantial profit. But most of 
this profit must be kept in the business; the monev is needed to restock 
the inventory, which must now be clone at the new high price level. 


National income analysts have sought ways to avoid including this profit, 
which they do not regard as real, in the total of national income, and 
they likewise wish to avoid writing down the national income in periods 
of falling prices when large nominal losses are recorded under this tradi- 
tional method of inventory accounting. 

On this point, great progress has been made in the United States in 
the past decade, owing largely to the pioneering work of Kuznets and his 
associates, and to the later work of the Department of Commerce. The 
British have been unable to get far with this problem, since they lack 
adequate data on capital formation in general, and inventory changes in 
particular. The method used by Kuznets cannot readily be summarized 
here, but the effect is to obtain a measure of the changes in inventory 
holdings during a given year valued at the price level of that year, hence 
uninfluenced by the particular cost level that happened to obtain when 
the inventory was accumulated. Of course, there still remains the prob- 
lem of making this years figure comparable with that of another year 
when a different current price level obtained, but this is part of the gen- 
eral problem of adjusting the national income total for year-to-year and 
place-to-place comparisons. 

Progress has also been made by Kuznets, using Fabricant's computa- 
tions, in adjusting depreciation charges to the price level of the year for 
which the charges are made. 

Financial Intermediaries 

When a financial intermediary, like a commercial bank, renders serv- 
ices free of direct charge, or at a service charge set below cost, there arises 
the possibility that a part of the economy's total of final goods and services 
will be understated through the omission of part or all of these services. 
The financial institution counts on covering part of its expenses by its 
receipt of income from investments that are made with funds belonging 
not to it but to its customers. Although the treatment of financial inter- 
mediaries had been discussed for some time, the analysis had been chiefly 
in terms of these intermediaries as aggregations of individuals, and it re- 
mained for Yntema, of the Department of Commerce, to discover and 
point out clearly the problem of imputed income that was involved. As a 
result, the revised Commerce series marks a notable advance in the treat- 
ment of the income of financial intermediaries. To the depositors of the 
bank there is imputed an amount of interest equal to the difference be- 
tween the amount the bank receives as interest and dividends and what it 
pays the depositors as interest. The imputed interest is allocated among 
business firms, governmental units, and individuals. Each of these is then 


considered as spending this imputed interest in purchasing services from 
the bank. 

III. Advances in the Compilation and Publication 
of National Income Data 

The advances made during the past ten years in the compilation and 
publication of national income data have been very great indeed. Data 
are now at hand in some degree for several countries for which virtually 
nothing was available a decade ago. And in those countries that were 
already publishing substantial amounts of data in 1937, notably the 
United States and Great Britain, the improvement in range, precision, 
and tabular presentation of the data has been remarkable. 

The countries for which little or no data were available ten years ago 
may be grouped into three classes— those for which that condition still 
obtains; those for which there have been computed some rough estimates 
of total, with occasionally some of the sub-totals specified; and those 
which have published fairly detailed series covering a period of several 
years. The present essay cannot give a complete statement of progress 
for each country with respect to the national income data that are in 
preparation or that have been completed, but a few major examples will 
illustrate the varying degrees in which progress has been made. 

Canada, Australia, Eire, and the Netherlands have recently published 
detailed series for several years. The Canadian data, for example, include 
the period 1938-46. They are published as "National Accounts: Income 
and Expenditure," compiled by the National Income Unit of the Domin- 
ion Bureau of Statistics. The factor-payment side of the compilation is 
not so detailed as in the British and United States estimates; rent, inter- 
est, and corporate profits are lumped in the item "investment income." 
On the product side, domestic investment is divided between plant and 
equipment, and inventories (a division which the British White Paper 
has not yet achieved), but the data on consumer outlay are not divided 
between durable goods, semi-durable goods, and services, as in the United 
States series. The Canadian figures also include a series on personal in- 
come. There is no statement of the saving-investment equality. Totals 
are given for both national income (at factor cost) and gross national 

For France, the Commissariat General du Plan de Modernisation ei 
d'Equipement has published estimates of national income in considerable 
detail, but the data cover only the two years 1938 and 1946 (with some 
totals for 1929, and projections for 1947 and 1950). 


In Germany it appears that little or no progress has been made, either 
by the economics staffs of the occupying forces or by the Germans them- 
selves, in developing current national income data that would match in 
scope and detail the German series of the prewar years. 

The progress made in the United States during the past ten years may 
be noted by comparing the Department of Commerce booklet of Novem- 
ber 1938, Income in the United States, 1929-37, with its revised series in 
the July 1 947 Supplement to the Survey of Current Business, "National 
Income and Product Statistics of the United States, 1929-46." The most 
striking improvement is, of course, the computation of the national in- 
come (or, rather, the gross national product) from the product side. The 
earlier publication contained no information on how the national income, 
viewed as a flow of final goods and services, is divided among consumers' 
goods and services, private domestic investment in buildings, plant, 
equipment, and additions to inventory, net foreign investment, and gov- 
ernment goods and services. The product computation is made from sta- 
tistical sources different from those used in reaching the factor-payment 
total, and thus the two totals serve as a partial check on each other. Rec- 
onciliation of the two requires a small item of "statistical discrepancy/' 
The current series presents not only the totals for each of these items for 
each of the years 1929-46 (and in Kuznets' series the data are carried 
back to 1 91 9 on an annual basis), but also the sub-totals for durable 
goods, non-durable goods, and services to consumers; and sub-totals for 
federal war, federal non-war, and state and local purchases of goods and 
services. In the same category of improvement in data may be placed the 
Commerce tables showing the relation of the product total, when defined 
as gross national product, to the national income total as the sum of factor 

A second major improvement is the compilation of data on saving, and 
the relation of those totals to the total of investment (see the sub-section 
above on "Saving and Investment")- These data are brought together, on 
a gross basis, in a table of "Gross Savings and Investment Account." At 
this point attention should be called to the work done by the Securities 
and Exchange Commission in estimating liquid savings of individuals. 
This series is especially significant for comparison with the Commerce 
data, since it is compiled from different sources. Conceptual differences 
in coverage make precise comparison difficult, but the July Supplement 
presents the summary results of such an inquiry. 

Another major advance over the 1938 compilation is the construction 
of double-entry tables for each of five sectors in the economy, covering 
both the factor-payment and the product items. The total of wages and 


salaries of employees of private business concerns, for example, is entered 
on the left-hand side of a "consolidated business income and product ac- 
count," and on the right-hand side of a "personal income and expenditure 
account." The other three accounts are: "consolidated government re- 
ceipts and expenditures account," a "gross savings and investment ac- 
count," and a "rest of the world account." These double-entry accounts 
as presented in the July Supplement cover the year 1939. A highly con- 
densed double-entry table entitled "The Nation's Budget" was published 
in the 1946 and the 1947 Budget Messages of the President, and appears 
currently in the President's Economic Report. r 

A fourth important achievement of the Department of Commerce is 
the speed with which it makes available its national income and product 
data following the close of the year, and the issuance of quarterly esti- 
mates, soon after the close of each quarter, of national income totals at 
annual rates adjusted for seasonal variation. 

Equally notable progress in the assembly and publication of national 
income data in the United States has been shown in the series compiled 
by Kuznets and his associates at the National Bureau of Economic Re- 
search. While the Department of Commerce has taken over the task of 
maintaining an annual current series, and has limited its data to 1929 and 
the following years, Kuznets has been engaged in refining his annual 
series covering 1919-38, and pushing the product data as far back as 
1 869, in terms of averages for a decade for the earlier period. One result 
of this activity is the publication of by far the most detailed series on 
national income that has yet been made public for any country (except- 
ing only the Swedish estimates noted earlier in this article): National 
Income and Its Composition, 1919-38. This volume sets a standard for 
explicit statement of the sources utilized, the reasons for decisions taken 
in combining the underlying data in one way rather than another, and 
the publication of the many industry-by-industry sub-series. 

The British White Papers on national income represent an advance 
over the earlier pioneering work of Stamp, Bowlev, and Colin Clark 
chiefly in the systematic presentation of the material in the light of re- 
cent developments in economic theory, especially with regard to the rela- 
tions of the government accounts to the rest of the economy, and the 
saving-investment equality. There has also, of course, been a growth in 
the amount of data available. The successive White Papers have varied 
considerably in the amount and type of data they have presented, as well 
as in the amount of explanatory material accompanying the data, and 
some of the most useful or interesting arrangements and explanations of 
an earlier issue have not been repeated in a later one. Taken as a whole, 


the scries represents great progress in the compilation and presentation or 
national income data, but the student specializing in national income will 
want to be sure that he has inspected all the issues, not merely the most 
recent one. 

With respect to the availability of data suitable for compilation in na- 
tional income estimates, the decade has seen both advances and reverses. 
In the United States, salary and wage data from the Federal Bureau of 
Old-Age and Survivors* Insurance, the state unemployment compensa- 
tion agencies, and the Federal Railroad Retirement Board became avail- 
able just in time (1940) to help offset the handicap imposed by the fact 
that during the defense and war periods the biennial Census of Manu- 
factures was discontinued. The first Census of Manufactures since that 
for 1939 will be taken in 1948, covering the year 1947. The data com- 
piled from income tax returns and published annually in Statistics of 
Income, by the Federal Treasury, are presented in more detail and with 
less time lag than a decade ago. On the other hand, Congress has shown 
itself unreceptive, in recent years, to pleas for funds for various types of 
census in addition to that for manufactures, for studies of the distribution 
of consumer income, and for similar data. The economy wave in the ap- 
propriations passed in 1947 struck the Census and other fact-gathering 
agencies severe blows, and the effect will be felt for some time to come in 
less adequate national income data than had been hoped for. The present 
writer has the impression that the outlook is not much better abroad. The 
British data on capital formation are inadequate, but the government, al- 
though obviously interested in planning, seems unwilling to push for- 
ward with censuses; many income tax data are not made available, in 
summary form, even where they are known to exist, and the Board of 
Inland Revenue so far shows little indication of changing its policy of 
withholding much of such information. In more than one Western Euro- 
pean country the government has ordered the production of national 
income estimates without providing for the necessary basic censuses. It 
almost seems that the very ingenuity of the national income estimators in 
constructing bricks with very little straw may prove self-defeating; gov- 
ernment legislators, if not government officials, like to believe that a na- 
tional income estimate is something that can be produced without much 
expense. They may even come to believe that it is not an estimate but a 
recorded fact. 

The distribution of families and single persons by size of income is one 
of the sectors of the national income field in which notable progress has 
been made in the United States during the past decade. The great survey, 
under the auspices of the National Resources Committee, of income tor 


the twelve-month period ending June 30, 1936, which included personal 
interviews with 300,000 families, was published in 1938. The one other 
similar study covering the United States was that for the calendar year 
1 94 1 and the first quarter of 1942, published in 1945. The decade also 
saw the completion and the appearance of the first volumes of three state- 
wide studies of income distribution, for Delaware, Minnesota, and Wis- 
consin. The subject has been explored in some particular phases in two 
of the volumes in the National Bureau's "Studies in Income and Wealth" 
series. Sample surveys of distribution by income size have been obtained 
from time to time in connection with the monthly Labor Force surveys of 
the Census Bureau and the liquid-asset surveys made under the auspices 
of the Federal Reserve Board. A considerable amount of information was 
obtained in the last decennial census of population. 

There is nothing available for any other country to compare with these 
surveys, even though one of their major effects is to stimulate an appetite 
for more and to lead to a realization of how much we should like to know 
and cannot know until adequate continuing financial support for such 
projects becomes available. Similar comments apply to the accompanying 
studies on the use consumers make of their incomes. 

Other areas in the United States in which substantial improvements 
have been made in the basic data include rents received by individuals, 
where the Department of Commerce has been able to utilize data from 
Statistics of Income to a greater degree than before; income from profes- 
sional activities, which has been thoroughly studied by Milton Friedman 
and Simon Kuznets, and which Commerce has made the subject of ex- 
tended inquiry by questionnaires; and consumer outlay, which, instead of 
being estimated as a residual, by subtraction of all other kinds of product 
from the total of products (itself derived from the total reached by add- 
ing the factor payments), is now estimated with the aid of data on output 
at the manufacturing level, with allowance for wholesale and retail 
mark-ups, etc. This is not a complete list, but it indicates the diversity of 
fields in which it has been found possible to improve the data despite 
the relative lack of support from Congress. 

IV. Uses Made of National Income Data 

A summary of the uses made of national income data during the past 
decade must be kept within fairly narrow bounds if it is not to become a 
summary of activities in fields of economic analysis that are covered in 
other monographs in this volume. The concepts and data of national 
income are tools, and they have been found useful in several sectors. 


Here, only a brief indication can be given of what these sectors are and 
how the tools have been employed. 

With respect to government finance, the inflationary or deflationary 
pressures in existence at the moment have been described in recent years 
in terms of inflationary or deflationary "gaps." These gaps are simply in- 
consistencies of certain sub-totals within the national income or gross 
national product totals as assumed in the first stage of the analysis. A pre- 
liminary assumption of an unchanged price level, for instance, is made, 
and the amount that consumers would want to spend, out of the assumed 
total income, is compared with the amount of consumer goods that would 
be available in view of the total amount of production and the total 
amount of capital-goods production, exports, and government purchases, 
that are being assumed. Or the amount that could be expected to be saved 
is compared with the expected total of investment. The inconsistencies 
are then removed by altering the basic assumptions about the price level 
or the total amount to be produced or the amount of production that is to 
be in the form of non-consumer goods. 

This technique is at least useful in exposing hidden inconsistencies in 
any forecast. Fiscal policy prescriptions can be tested for the presence of 
quantitative absurdities by casting them in terms of this national income 
analysis. This is not to say that the mechanism can be guaranteed to 
detect all major errors in the forecast, or that it can in some way lead 
positively to the correct forecast. The failure of more than one forecast in 
recent years, based on this analytical framework, is evidence that, at the 
moment, the technique has more value in eliminating from discussion a 
number of erroneous conclusions than it has in pointing directly to the 
right one. But even this is a great advance in the analysis of government 
finance. A comparison of the kind of information and analysis, or rather 
the lack of it, that was available for formulating federal financial policy 
in World War I, and in the great deflation of the early logo's, with that 
which was available, and published (in Congressional Hearings) in 
World War II sufficiently demonstrates the point. 

The topic covered in the immediately preceding paragraph is of course 
but part of a much larger one, namely, the attempt to forecast the level of 
business activity for months or years ahead, not with particular regard to 
fiscal policy, but in general. In this attempt, as well as in the narrower 
one of estimating the pressures that would be exerted by changes in fiscal 
policy, a considerable amount of research has been done in constructing 
probable relationships between various sub-totals of the national income. 
For example, the amount that consumers may be expected to try to spend 
out of an assumed disposable income (personal income after taxes) is esti- 


mated on the basis of past relationships of these two amounts. The rec- 
ords are of too short duration, and the years they cover are too extraordi- 
nary (especially the first half of the 1930*5, and the war years), to inspire 
a great deal of confidence in the resulting ratios as predictors, but again, 
they may at least dispel erroneous ideas that would otherwise get into the 
analysis. Moreover, their value may well turn out to be greater than the 
tone of the present comment would suggest. The techniques used in de- 
veloping these ratios and adapting them for predictive purposes fall out- 
side the scope of this paper. But in general it seems that the work is 
emerging from an early stage, into a more sophisticated period. In the 
early years of pioneering enthusiasm, every bit of national income data 
was eagerly seized upon for use in constructing a predictive mechanism, 
sometimes without due regard to the degree of error in the historical sub- 
totals (which, after all, are only estimates themselves, for the most part) 
or even without a full understanding of the conceptual content of some 
of the sub-totals. That pioneering enthusiasm, somewhat uncritical but 
probably indispensable to an opening up of this new and difficult terri- 
tory, is being supplemented by a more methodical, if slower, approach, 
from which it seems reasonable to expect important developments. The 
revised data issued in 1947 by the Department of Commerce for the 
period 1929-46 will aid greatly in the refinement of this analysis. 

The annual totals of national income, or gross national product, have 
been used to establish trends in other annual data. For example, the num- 
ber of non-agricultural and non-professional business firms in operation 
in the American economy in 1929-40 has been correlated with the gross 
national product (excluding government and agriculture), and the post- 
war data on number of firms is observed to show a close return to the 
prewar relationship, after the large decrease in number of firms caused by 
the war. 3 

The national income data have been especially useful in giving a cor- 
rect sense of proportion about the structure of the economy, entirely apart 
from any attempts at prediction. The amount, and proportion, of the 
nation's output that is going into capital equipment and foreign invest- 
ment, and hence is not available for immediate use by consumers or gov- 
ernment, is indicated by the national income data; and estimates by 
Kuznets as far back as 1 869 give at least a rough impression of changes 
that have occurred in this proportion over the decades. As a result of the 
compilations published during the past ten years much more is known 
about the proportion of output that takes the form of investment, and of 
the proportions in which this sub-total is divided among plant and equip- 

8 Survey of Current Business, July 1947, XXVII, p. 15. 


ment expenditures, construction, increases in inventory, and foreign 
investment, and, in some series, government capital outlay. The interpre- 
tation of these findings has progressed rapidly in certain directions, nota- 
bly, as already indicated above, in illustrating and extending Keynesian 
theory. At the same time, it appears that much, if not the main amount, 
of interpretation of the data on amount of investment is yet to come. Not 
as much progress has been made as might have been expected in com- 
paring and contrasting the amounts of capital formation of different 
types, or in drawing inferences from the absolute and percentage amounts 
of total investment. 

The national income data are also now the common mode of expressing 
the size of the economic role that government plays in the economy : the 
ratio of taxes to national income, or of government expenditures to net or 
gross national product. These ratios indicate the importance of the gov- 
ernment in its role of dispenser of goods and services. Some of this prod- 
uct, however, has been produced in private business establishments, not 
in government departments. The cost of fire protection dispensed free of 
charge by the government is made up, in part, of the firemen's salaries, 
and, in part, of the cost of the fire trucks, hose, etc. The firemen's salaries 
represent production directly by the government; the trucks and hose are 
produced by private business firms. The role of the government as a pro- 
ducer is shown by comparing the total of factor payments made by the 
government with the total of factor payments in the economy. 

Some important decisions on the definition of goods and services pur- 
chased by government, and on factor payments made by government, 
have been taken recently without as much discussion in the journals or 
proceedings of conferences as could be justified. The decision of the 
Department of Commerce to omit from national income all interest pay- 
ments by government is a case in point. The decision was not hastily 
made; long discussions preceded it; but relatively little of the debate has 
appeared in print where it could be analyzed by others. In this respect the 
advance in the use or at least in the understanding of government data in 
the national income has not been as great as could have been wished for. 
A similar comment may apply to the recent decision of the Department 
of Commerce to define, as a factor payment, corporate profits before de- 
ducting corporation income tax rather than corporate profits after deduct- 
ing that tax, as had been the practice theretofore. The treatment of the 
corporation income tax in defining the factor payment of profits raises 
some fundamental and difficult problems of the definition of a factor pay- 
ment and, indeed, the whole idea of a system of distribution of economic 
rewards. Traditional economic theory has not adequately re-analyzed the 


concept of a factor payment in the light of the existence of income and 
other taxes, and a discussion linking up national income theory and eco- 
nomic theory at this point seems somewhat overdue. Again and again the 
reader in the national income field encounters these cases where the deci- 
sions made in the compilations of the data have to some extent outrun the 
relevant abstract analysis available in published form. This remark is not 
a criticism of the compilers; they have had to make the decisions, if the 
data were to appear at all, and in their oral discussions and office memo- 
randa much abstract analysis has doubtless been used in reaching the 
decision. Moreover, the Department of Commerce staff has made an 
effort to obtain the benefits of theoretical criticisms from others. The im- 
plication intended here is that, in one way or another, the national in- 
come specialist and the student of economic theory need to be drawn 
together more, and to take more interest in each other's problems. 

In war time the study of the product components of national income 
becomes particularly important. A "maximum" war effort is, of course, a 
relative concept, but even a relative maximum cannot be aimed for very 
well unless data are at hand on the total past output of the economy, the 
total possible output under assumed wartime conditions respecting the 
labor force, capital equipment, etc., and the minimum amount that must 
be allowed for continued production of consumer goods and non-war 
capital formation (perhaps negative) and non-war government services. 
The maximum output minus this minimum sub-total gives at least a goal 
to aim for in the production of war goods and services. In the absence of 
such data it seems likely that the aim will be set below rather than above 
this maximum. The data on consumer goods and services produced 
during 1941-45 indicate that the United States fell considerably short of 
a maximum practicable war effort; yet, would the President have dared to 
set the sights on war production as high as he did if data on national 
income and its components had not been available? 

And in aiming at the maximum, the national income sub-totals on the 
factor-payment side indicate what kind of behavior is expected of con- 
sumers with respect to departure from established habits of spending and 
saving; the proportion of personal disposable income that must be saved 
if inflation is to be avoided can be computed, approximately, and this 
figure in turn gives some indication of the extent and severity of direct 
controls, like rationing, that must be employed in view of the amount of 
taxation assumed (which is the chief element in determining how much 
of the consumers' income will be disposable income). 

It must not be inferred that these wartime uses of the national income 
data always reached a high degree of precision. For some purposes the 


figures could come out a few billions one way or another without appre- 
ciably changing the conclusions that would be drawn from them. But the 
sub-totals, or most of them, were so magnified under the pressure of the 
war economy that it was helpful, indeed essential, to have even an answer 
that might be several billions in error. 

The past decade witnessed the first use, as far as the present writer is 
aware, of national income data as determinants of the amount of aid that 
should be extended by a government to its political subdivisions; more- 
over, national income data were used in determining how an interna- 
tional financial burden should be apportioned. 

Federal aid to the states is determined in part by the per capita incomes 
of the several states, in grants under the Hospital Survey and Construc- 
tion Act of 1946, and in the School Lunch Act of the same year. The 
data utilized are the Department of Commerce series on income pay- 
ments to individuals, and hence do not include undistributed corporate 
profits. This omission needs to be kept in mind in deciding how much 
weight to give to the state income-payments data as determinants in the 
distribution of federal aid, as does also the unavoidable incompleteness in 
the computations of dividend and other property income on the basis of 
the state in which the recipient resides. 

The international use of national income data was in the allocation of 
the costs of the United Nations Relief and Rehabilitation Administra- 
tion. A requisition was made of each nation equal, with important excep- 
tions, to 1 per cent of the national income. In view of the rudimentary 
nature of the estimates of national income for most of the countries, this 
first attempt does not supply much useful experience for future calcula- 
tions of the same kind, but the fact that national income was considered 
to be a possible international "tax" base is another indication of the inter- 
est that national income analysis will have for students of public finance. 

The data on individual income, arranged by size classes, and its alloca- 
tion between saving, spending, and tax payments, have been the subject 
of considerable analysis during the past few years. For the most part, the 
analysis has been concerned with discovering and stating clearly some of 
the more interesting or puzzling relationships revealed by the data rather 
than with searching for the reasons for those relationships. The large 
amount of data at hand (inadequate though it still is for many purposes) 
from studies differing in time, scope, and geographic coverage have made 
it necessary to spend much time and effort on this preliminary surveying 
of the material, including attempts to reconcile data from different 
studies; no doubt the coming decade will see somewhat more of interpre- 
tation of the relationships. 


Two of the most interesting uses of national income data are those 
made by Wassily W. Leontief and Morris A. Copeland. Leontief, de- 
pending largely on Census of Manufactures data, has computed the 
quantitative input and output relationships of each major sector of the 
economy (automobiles, petroleum and natural gas, yarn and cloth, agri- 
culture, etc.) with each of the other sectors. Copeland is at present en- 
gaged in compiling data on the money flows within the economy, a 
project which of course extends beyond the boundaries of national in- 
come analysis. 

Since changes in fer capita national income, as distinguished from any 
one of its constituent items, are a partial indicator of changes in the mate- 
rial welfare of the group of persons making up the nation, it might be 
supposed that one of the most prominent uses made of the national in- 
come figures would be the drawing of inferences regarding the advance 
or decline in material welfare over a period of years, or, as among differ- 
ent nations, differences in welfare at any one time. 

Such use has in fact been made of the data, but not perhaps to as great 
a degree as might have been expected. Moreover, the inferences drawn 
have been for those kinds of time-span that might have been thought 
most difficult to analyze: decades in the past, including the latter half of 
the nineteenth century, and decades projecting into the future. There 
has been relatively little year-to-year comparison for the recent past. 
Totals have indeed been deflated by product-price indexes, and the reader 
is thus enabled to draw his own conclusions about the fluctuating totals in 
the 1930's, for instance; nor have the implications of these fluctuations 
passed entirely unnoticed. But there has not been much intensive analy- 
sis or extensive debate over whether 1932, for example, was a year of less 
material welfare than 1933. It may seem self-evident that so refined an 
analysis cannot or should not be undertaken, but in any event the relative 
emphasis on long periods in the past and in the future is another indica- 
tion of how the theory of national income totals, including the theory of 
deflating for price changes, has suffered a relative lag. 


Very little has been accomplished during the decade in examining and 
refining the conceptual problems of national wealth. Perhaps this is be- 
cause continuing data have been at hand for constructing national income 
estimates, whereas little has been done in recent years to collect data for 
the purpose of computing national wealth. Still it might be thought that 
the discussion of concepts would not be so entirely dependent for stimu- 
lation upon the existence of current data. The question is probably not 


so much one of stimulation in an absolute sense, as it is one of relative 
stimulation. The task of constructing national income estimates has 
seemed so urgent and at the same time so practicable that it has easily 
absorbed all the energy available from those working in the field of in- 
come and wealth. 

In the process of refining the concepts useful for income analysis, there 
has, of course, been considerable implicit or fragmentary progress in the 
same task in the field of national wealth; this will be clearly apparent 
when that latter task is explicitly approached in a comprehensive study. 
Inventory valuations, depreciation charges, the distinction between con- 
sumer outlay and investment— these and many other subjects on which 
progress has been made in the past ten years would have had to be 
worked up also in any study of national wealth. 


The foregoing essay has purposely refrained from citing particular 
works or articles, except as reference is made to them in the text of the 
discussion; the attempt has been to give a general view of the broad ad- 
vances in national income theory and compilation. The following sugges- 
tions are offered in the hope that they may be useful to those who, while 
not specializing in this field, would like to study some of the more recent 
comprehensive works. In addition to the particular works cited here, the 
reader is referred to the series of volumes of proceedings (Studies) of the 
Conference on Research in National Income and Wealth. 

Barger, Harold, "Outlay and Income in the United States, 1 921 -193 8," Studies in 
Income and Wealth, IV (New York, 1942), pp. xxvii + 391. 

Clark, Colin, The Conditions of Economic Progress (London, 1940), pp. xii ■+■ 

Derksen, J. P. D., A System of National Book-keeping; Illustrated by the Experi- 
ence of the Netherlands Economy. National Institute of Economic and Social 
Research. Occasional Papers, X (Cambridge, England, 1946), pp. 31. 

Fabricant, Solomon, Capital Consumption and Adjustment (New York, 1938), 
pp. xx + 271. 

Friedman, Milton, and Kuznets, Simon, Income from Independent Professional 
Practice (New York, 1945), pp. xxxiii + 599. 

Gilbert, Milton, and Jaszi, George, "National Product and Income Statistics as an 
Aid in Economic Problems," reprinted in William Fellner and Bernard F. Haley, 
eds., Readings in the Theory of Income Distribution (Philadelphia, 1946), pp. 

Hicks, J. R., and Hart, A. G., The Social Framework of the American Economy: 

An Introduction to Economics (New York, 1945), pp. xvi + 261. 
Kuznets, Simon, National Income: A Summary of Findings (New York, 1946), 

pp. 144. 


, National Income and Its Composition, 1919— 1938, 2 vols. (New York, 

1 941), pp. xxx -+- 929. 

, National Product Since 1869 (New York, 1946), pp. xvi -f 239. 

Perroux, Francois, Uri, Pierre, and Marczewski, Jan, Le Revenu National (Paris, 
1947), pp. 310. 

Shoup, Carl S., Principles of National Income Analysis (Boston, 1947), pp. xv 
+ 400. 

Stone, Richard, Definition and Measurement of the National Income and Related 
Totals. Memorandum Submitted to Sub-Committee on National Income Statis- 
tics of the Committee of Statistical Experts, League of Nations (Princeton, June 
1946), pp. 23-128 of mimeographed report of the sub-committee. 


United States: Department of Commerce, National Income Division, "National 
Income and Product Statistics of the United States, 1929—46." Supplement to 
Survey of Current Business, July 1947, XXVII, pp. 54. 

United States: National Resources Committee, Consumer Expenditures in the 
United States: Estimates for 1935-36 (Washington, 1939), pp. ix + 195. 

: Consumer Incomes in the United States: Their Distribution in 1935—36 

(Washington, 1938), pp. viii + 104. 

Great Britain: Financial Secretary to the Treasury, National Income and Expendi- 
ture of the United Kingdom, 1938 to 1946. Cmd. 7099 (London, 1947), 
pp. 60. 


Hicks, J. R., "The Valuation of the Social Income," Economica, May 1940, New 
Series VII, pp. 105-24. 

Hoffenberg, Marvin, with assistance from Mabel S. Lewis, under the direction of 
John H. G. Pierson, "Estimates of National Output, Distributed Income, Con- 
sumer Spending, Saving, and Capital Formation," Review of Economic Statis- 
tics, May 1943, XXV, pp. 1 01-174. 

MacGregor, D. C, "Recent Studies on National Income," Canadian Journal of 
Economics and Political Science, February 1945, XI, pp. 115— 129; and May 
1945, XI, pp. 270-280. 

Schwartz, Charles F., and Graham, Robert E., Jr., "State Income Payments in 
1946," Survey of Current Business, August 1947, XXVII, pp. 9-24. 

Stone, Richard, "The Measurement of National Income and Expenditure: A 
Review of the Official Estimates of Five Countries," Economic Journal, Septem- 
ber 1947, LVII, pp. 272-298. 


Henry H. Villard 

I. Introduction 

In the social sciences our accumulated knowledge is so small and the 
unexplored areas so vast that of necessity we measure progress by the 
understanding we obtain of particular and pressing problems. Thus 
Adam Smith did not write primarily as a scholar but rather as a social 
surgeon to remove from the body politic the surviving malignant remnant 
of Mercantilism, while the emphasis of Ricardo and Mill on diminishing 
returns and the rent of land directly reflected their interest in the ulti- 
mately successful campaign of the rising business classes to end the im- 
pediment to further industrialization represented by the Corn Laws. 

In the same way recent monetary theory directly reflects the unprece- 
dented depression which rocked the industrialized world during the nine- 
teen-thirties; in the United States, perhaps worse hit than any other 
country, the increase in productive capital, which had averaged 6 per 
cent a year for the first three decades of the century, over the 'thirties as 
a whole was negligible in amount. As a result the center of interest has 
in general shifted from the factors determining the quantity of money 
and its effect on the general level of prices to those determining the level 
of output and employment. In addition, the purely monetary devices for 
control, on which great store had been laid, were found to be broadly 
ineffective, taken by themselves, in bringing about recovery from the 
Great Depression. And finally, as a result of the way in which the war 
was financed, it seems quite likely that it will prove impossible to use 
such devices for the effective control of a future boom. The general 
change in emphasis is well indicated by the altered character of univer- 
sity courses: in 1930 an outstanding elementary text devoted 144 of its 
1250 pages to Money and Banking and 16 to the Business Cycle; in 1947 
a new elementary text devoted 205 of its 700 pages to National Income 
and Employment and 55 to Money and the Interest Rate! 

The implications of this decreased emphasis need to be made quite 
clear. Basically it reflects a reduced interest in the factors influencing the 
quantity of the available means of payment and an increased interest in 



the factors influencing the level of spending. Apart from unguarded 
statements monetary theorists have of course generally been aware that 
money had not only to be created but also spent if it was to have any 
effect on the economy. But up until relatively recently the economy ab- 
horred any large amount of idle balances. Thus the emphasis of monetary 
theorists was on changes in the quantity of money, accompanied by the 
sometimes stated and sometimes implied assumption that balances, once 
created, would not long stay idle. It is with monetary theory in this nar- 
row sense that this paper will be mainly concerned. 

This narrow construction of monetary theory perhaps requires defense. 
There can be no question that the fields covered by monetary theory, the 
theory of compensatory fiscal action, and business cycle theory are at least 
closely related if not actually overlapping. Further, business cycle theory 
to a major extent, and fiscal theory to a lesser extent, evolved out of mone- 
tary theory; as a result courses and economists have often in recent years 
been labeled "monetary" when in fact they were dealing with employ- 
ment, output, and income. But a broad use of monetary theory would not 
only make it a synonym for business cycle theory but also would make it 
impossible to describe separately work dealing predominantly with the 
factors influencing the quantity of money. In other words, it has seemed 
desirable to separate monetary from fiscal and cyclical theory in such a 
way as to minimize the overlapping between the fields. I sincerely hope 
that those who have been accustomed to define monetary theory more 
broadly will not take offense at the relatively limited meaning which is 
used in what follows, and will remember that it is monetary theory in the 
narrow sense which is described as having declined considerably in im- 
portance in recent years. It should hardly be necessary to point out that 
the depression greatly increased the importance of monetary, fiscal, and 
cycle theory taken together. 

This paper will start with a description of some of the recent changes 
in environment, both to summarize the contribution of those who have 
worked on the statistical side of monetary problems and at the same time 
to explain why the changes involved have been so largely responsible for 
our decreased interest in the quantity equation in recent years. A second 
section will be devoted to the concepts of monetary equilibrium which 
were developed as alternatives to the quantity equation, while a third 
will cover changes in the explanation of the determination of the rate of 
interest. Two final sections will deal with the financing of the war and 
the heritage that the war has left us. 

Any summary of developments during such a dynamic period in mone- 
tary thought cannot fail to be impressionistic. In a literal sense, therefore, 


the summary will inevitably be wrong; but in a broader sense it is only 
the surviving impact of thought which is important. The obscure and 
forgotten passage or the uses to which discarded tools of analysis might 
have been put are not what matter, however much they may delight the 
historian and prove that there is nothing new under the sun. It is, then, 
to the broad view that this essay will be devoted; he who seeks details 
will have to look elsewhere. 

II. The Changing Monetary Environment and the Decline 
of the Quantity Equation 

Few analytical devices in economics have been as useful over as long a 
period as the quantity equation of exchange. By the start of the logo's 
there was considerable agreement that the equation was perfectly valid 
when properly— i.e., tautologically— defined. 1 As first propounded by Pro- 
fessor Fisher, T included "all things sold for money" during any period, 
so that V became all uses of money to buy "things" and P a hybrid price 
level applying to all sales of "things" for money. The implications of this 
usage were not made clearer by a general tendency to refer to T in this 
sense as "trade" and to P as the "general" price level. Even if there can 
be no analytical objection to this formulation, when an attempt is made 
to derive statistical values, especially from figures for bank debits, many 
problems arise. Even today little is known regarding the extent to which 
bank debits reflect "money to money" transactions, such as transfers of 
funds from one account to another of the same economic unit; while the 
inclusion of sales of not only current output but also such diverse things 
as stocks and bonds, urban and agricultural land, and second-hand cars 
and antique furniture— to mention only a few examples— makes it diffi- 
cult either to calculate an appropriate price index for P or to attach any 
significance to the resulting level of T. 

Although somewhat less easy to understand and therefore less gener- 
ally known, the "cash balance" version of the quantity equation (espe- 
cially as it was used at Cambridge University) and other examples of the 
income approach were more in line with recent developments. 2 This does 
not result from the cash balance equation itself— most simply written as 
M equals &.PT where k equals i/V— as the formulation is subject to the 
same problems and criticisms as the Fisher equation if the various terms 

1 If V is the use of M to buy T, T specific items sold for M, and P prices of T when 
sold for M, then the equation is valid because it is a truism. The charge that the equa- 
tion was not valid arose because in some earlier presentations use had been made of such 
approximations as existing price indexes, which rendered the equation formally incorrect. 

2 For the derivations of these approaches, see A. W. Marget, The Theory of Prices 
(New York, 1938), pp. 302-343 and 414-458. 


are given the same meaning. But the emphasis was on k defined as the 
relationship between people's money balances and their incomes, so that 
PT referred, not to the total value of monetary transactions, but to the 
total value of transactions in current output— in other words, the national 
income. This relationship seems clearly more significant for business 
cycle problems than Fisher's V. 3 

It was a combination of this shift in theoretical interest from the "trans- 
actions" to the more fruitful "income" version of the quantity equation 
and the availability of national income estimates that made substantial 
statistical progress possible during the logo's, while the statistical wort 
itself stimulated further analysis of the variables involved. Thus relatively 
full information regarding the behavior of the terms of the quantity 
equation did not become available until the controversy over the "quan- 
tity theory" as an explanation of prices had largely died down. 4 

Perhaps the best place to start a description of the statistical progress 
which took place is with the clarification— largely by Currie and Angell— 
of the concept of "money," which has come increasingly to mean cur- 
rency outside the banking system (in the hands of the public) and 
demand deposits (deposits subject to check), including all government 
deposits but excluding all interbank deposits. Currie uses this meaning 
exclusively, while Angell also computes values for "total" money (includ- 
ing time and saving deposits) but lays greatest emphasis on money in the 
first sense, which he calls "circulating" money. Although the term may 
of course be applied to any of a number of concepts, it seems clear that 
this definition is most useful when our interest centers upon the primary 
function of money as a means of payment or upon the most "liquid" form 
that assets can take. The emergence, moreover, of large holdings by indi- 
viduals of U. S. Savings Bonds, which are payable in a specific number 
of dollars on demand, has eliminated the claim of time and saving depos- 
its to special consideration. This development, plus the general accept- 
ance of the narrower meaning of money, has led to the development of a 
new concept of "liquid assets" to include money, time and saving de- 
posits, and U.S. Bonds, although there is of course no sharp line between 
assets which are or are not "liquid" but rather an infinite series of gra- 
dations. 5 

3 This point is discussed further on pp. 323-324 below. 

4 The volumes which contributed most to our understanding of the statistical magni- 
tudes involved were L. Currie, The Supply and Control of Money in the United States 
(Cambridge, Mass., 1934); J. W. Angell, The Behavior of Money (New York, 1936); 
and idem, Investment and Business Cycles (New York, 1941). Much interesting work 
has also been done by the Board of Governors of the Federal Reserve System and the 
Federal Deposit Insurance Corporation, particularly through the periodic surveys of de- 
posit ownership which are published in the Federal Reserve Bulletin. 

5 Estimates of the distribution among various holders of "liquid assets" in this sense 
have been published periodically in the Federal Reserve Bulletin since 1945. 


Probably the most striking fact to emerge from recent statistical studies 
is the constancy of income velocity before 1 929 and the size of the appar- 
ently permanent decline since that date. Between 1899 and 1929 income 
velocity— the national income divided by money as defined above— ranged 
from 2.72 to 3.35, a variation of less than 25 per cent; yet during the 
same period income and money increased more than fivefold. 6 

Income Velocity After 1929 


































2. 1 1 
































2. 1 1 




1 .81 













* In billions. Through 1939 the estimates are from Angell, 
loc. cit. Thereafter the total used is the sum of "Currency 
outside banks," "Demand deposits adjusted," and "United 
States Government deposits" for the end of June as re- 
ported in the Federal Reserve Bulletin. 

f In billions. All estimates are the revised national income 
series of the Department of Commerce. 

After 1929 velocity declined rapidly to about 60 per cent of its pre- 
depression level; thereafter, except for the boom year of 1937, it did not 
exceed 70 per cent of its previous level until the war. During the war 
velocity first rose to 2.50 during the relatively tight credit conditions 
which marked the start of large-scale war finance in 1942— a level higher 
than in any year since 1930— and then declined to 1.50 as the money 
supply expanded during the later years of the war. 7 Figures for the period 
since 1929 are given in the table above. 

6 J. W. Angell, Investment and Business Cycles, pp. 337-338. 

7 During 1 947 a decrease of 8 per cent in the supply of money (as a result of the 
reduction in Government balances to very low levels) brought about a rapid rise in 
velocity to a level close to that of the 'thirties. It will be interesting to see how long 
this increase continues. 


In addition to demonstrating the extent to which income velocity has 
declined, recent studies have clarified the factors influencing the maxi- 
mum level of income velocity. 8 Of basic importance are the intervals 
between successive payments during the circular flow of money from in- 
come recipients to producers and back to income recipients, the degree of 
overlapping of payment schedules (whether income received Friday is 
used to pay bills Saturday or vice versa), and the degree of business inte- 
gration; the amount of friction in the payment-transfer mechanism also 
plays a role in the result. Because it is impossible to eliminate "financial" 
transactions adequately, it is possible to determine only approximately 
how many exchanges take place in the course of the circuit flow of an 
average dollar from income recipient to producer and back again, but il 
appears that, as we are presently organized, roughly ten dollars' worth of 
"unfinished" output is exchanged for every dollar's worth bought by in- 
come recipients. 9 

The significance of these studies lies not so much in the actual numeri- 
cal estimates made as in their conclusion that at least minimum balances 
and maximum levels of income velocity are determined by relatively con- 
stant factors unlikely to change rapidly except under the impact of run- 
away inflations or drastic changes in payment practices or in the degree 
of business integration. It is true that at present and probably in the future 
actual balances will substantially exceed such -minimum balances, which 
reduces the practical importance of this information. But the understand- 
ing of the monetary process that has been achieved is considerable, as can 
be seen by comparison with formulations in which the level of actual bal- 
ances is taken as being determined by the quantity of "ready purchasing 
power" which people find it desirable "to keep by them." 10 For such a 
quantity would appear to be capable of rapid variation in any direction, 
when in fact a reduction in balances below the minimum level is likely 
to be difficult to achieve except under unusual circumstances. 

Finally, these statistical studies have made clear that stock market 
speculation can have a major effect on the level of exchange velocity 
without reducing the amount of money available for purchasing current 
output. Thus our broadest measure of exchange velocity indicates an in- 

8 Most of the credit again belongs to Professor Angell; see especially his article on 
"The Components of Circular Velocity of Money," Quarterly Journal of Economics, 
February 1937, LI, pp. 224-272. 

9 Idem, The Behavior of Money, p. 191. This includes "normal" financial transactions 
in which money is shifted between balances before being spent, but excludes transac- 
tions connected with the stock market and the like. Additional discussion will be found 
in H. H. Villard, Deficit Spending and the National Income (New York, 1941), p. 37, 

10 See Alfred Marshall, Money, Credit and Commerce (London, 1923), p. 43 ff. 


crease of 85 per cent from 1922 to 1929 at a time when income velocity 
was virtually constant. It was the distorting influence of changes of this 
sort— plus the difficulty of obtaining appropriate measures of the price 
level involved— which made the transactions version of the quantity equa- 
tion so difficult to interpret. 

To a major extent recent developments in monetary theory reflect the 
implications of these findings— especially the decline in income velocity. 
For up to 1930 the relative constancy of the relationship between money 
and income justified concentration on the factors determining the quan- 
tity of money— on monetary theory in the narrow sense which we employ. 
When changes in the quantity of money could be expected to have a 
broadly proportionate ultimate influence on incomes in either an upward 
or downward direction— despite year to year variations resulting from 
minor changes in velocity— it was natural to stress the importance of 
monetary changes, which appeared to be both strategic and controllable. 
Actually there probably still exists an upper limit to the expansion of the 
national money income that is possible with a given quantity of money; 
but, even if the limit is now somewhat lower than it was in the past, it 
seems highly probable that ever since 1929 the quantity of money in ex- 
istence could have supported a level of income substantially higher than 
that which actually prevailed. Hence it is the factors determining the use 
of a stock of money more than adequate to meet current or prospec- 
tive requirements that have become of primary importance in recent 

Furthermore, right up to the war, our production was less than that 
permitted by our labor force and plant facilities. In other words, the 
quantity of current output offered for sale, instead of increasing slowly 
along a secular trend line, was subject to wide fluctuations from one 
year to the next— sometimes with little or no variation in prices. Hence 
theoretical analysis has increasingly concentrated on setting forth both 
the determinants of the flow of spending and the effect of the resulting 
spending on output and employment, rather than the changes which the 
quantity of money could be expected to have on the level of prices under 
conditions in which it could be assumed that income velocity would be 
relatively constant and output at about the highest level permitted by our 
labor force and plant facilities. 

It should be made quite clear that the quantity equations are no less 
true for a period in which income velocity and output vary widely than 
for a period in which they are relatively constant; they are merely less 
useful. Although rarely if ever put forward in an unqualified form, the 
essence of the "quantity theory" was that a change in money could be 


expected to have a proportionate effect on prices, which is only true when 
there are no changes in either velocity or output. For the quantity equa- 
tions to be most useful, the conditions underlying the quantity theory 
must prevail. What is here suggested is that such conditions did in fact 
generally prevail before 1929— perhaps to a greater degree than was real- 
ized at the time. But when income velocity started to vary widely after 
1929, so that the quantity approach could only state that income would 
be equal to the money supply multiplied by a variable of unknown mag- 
nitude, other tools were developed to determine the level of incomes, 
although the quantity equations of course remained not only formally 
valid but useful for various purposes, especially elementary instruction. 
In the same way, the fluctuations in output that have taken place in 
recent years have made it far more difficult to determine the effect of 
changes in spending on prices; but here, in contrast to the previous case, 
relatively little has been done in the way of developing alternative tools 
of analysis. 

Keynes' A Treatise on Money 1 ' 1 is especially interesting in this connec- 
tion because it represents a transition from the monetary theory of the 
quantity equations to the modern theory of income, output, and employ- 
ment. In view of the emphasis on saving and investment contained in the 
Treatise, it is easy to forget that its "Fundamental Equations" summa- 
rized the factors determining the frice levels of consumption goods and 
output as a whole. Keynes started by distinguishing between the normal 
income of entrepreneurs— that "which, if they were open to make new 
bargains with all the factors of production at the currently prevailing 
rates of earnings, would leave them under no motive either to increase or 
to decrease their scale of operations" 12 — and their "windfall profits"— the 
difference between their actual receipts and their normal income. De- 
fined in this fashion, windfall profits (positive or negative) become the 
difference between the actual level of the national income and that level 
which would be just sufficient to continue the current level of output at 
the current level of factor costs— which may appropriately be called the 
"equilibrium" level. 

The broader and more important of the two Fundamental Equations, 
that determining the price level of output as a whole, was formulated by 
Keynes in the following fashion: 

n =*+!=* 

11 London, 1930. 
"Ibid., Vol. 1, p. 125. 


where n was the price level of output as a whole, E normal income ex- 
cluding windfall profits, O the volume of output, I the value of the cur- 
rent production of investment goods, and S saving out of normal income. 
Further, the Treatise makes clear that, under these definitions, windfall 
profits are equal to I — S. 13 Hence, as normal income plus windfall profits 
equals actual income, this equation in fact tells us that actual income 
divided by the volume of current output will give the price level of such 
output; or alternatively that the actual price level will differ from the 
"equilibrium" level by the excess of actual over "equilibrium" income 
divided by current output. 

To state that the price level of current output is determined by the 
actual expenditure on such output (i.e., the actual national income) di- 
vided by the volume of such output clearly does not represent an improve- 
ment of the quantity equations; hence the Treatise equations in reality 
must (and will below) be judged in terms of their contribution to con> 
cepts of monetary equilibrium. In other words, the analysis of the Trea- 
tise, although cast in the form of quantity equations, in a broad way 
marks the end of the equations as tools of new theoretical analysis in both 
Great Britain and the United States. This of course does not mean that 
the quantity equations were never heard of after 1930. Analytical and 
statistical work like that already summarized was largely inspired by the 
equations, while their place in elementary texts remained relatively se- 
cure, because of the ease with which they imparted a preliminary un- 
derstanding of monetary processes. But it seems fair to say that almost all 
new analytical work designed to explain the problems of the 'thirties rep- 
resented a break from, rather than an evolution of, the quantity equation 
approach, and made little use of quantity equation concepts. 

To this generalization there is a notable and outstanding exception. 
Professor Marget in his two-volume The Theory of Prices 1 * has sought 
both to defend the equations against the aspersions cast on them by ad- 
vocates of the newer approach (particularly Keynes) and also to renovate 
the equations as tools of theoretical analysis, particularly the transactions 
equation in the form originally propounded by Professor Fisher. There 
can be no question of Marget's contribution to the history of doctrine or 
of the extent to which he has demonstrated the falsity of many Keynesian 
generalizations, even when the generalizations remain suggestive. 15 But 
his contribution is considerably more than this; his painstaking discussion 

33 Ibid., p. 138. 

14 New York, 1938 and 1942. 

15 Keynes' suggestion that monetary and value theory have been inadequately inte- 
grated, for example, for me remains suggestive even after Marget has demonstrated that 
every economist since Aristotle applied the same methods of analysis to both fields. 


of the equations themselves has added insight into problems that must be 
faced in any careful formulation of the terms involved. To give but one 
example, his analysis of possible discrepancies between "current output" 
and 'goods sold" is undoubtedly definitive. 16 

Marget's plea for a return to the transactions type of equation, how- 
ever, is less likely to be accepted. The fundamental issue is a perennial 
one in economics: workability vs. completeness. It is Marget's position 
that any formulation that does not include all possible types of money 
and all possible uses is less than complete. Hence he objects to income 
velocity on the ground that it is a "hybrid" concept; for income velocity is 
actually an average relationship between all balances and the national 
income, most balances being in fact held against the purchase of "un- 
finished" output rather than against the purchase of the "finished" goods 
and services whose value adds up to the national income. Any such aver- 
age he rejects because it involves more than a simple relationship between 
cash balances and the specific outlay against which they are held, which 
is the only sort of relationship sufficiently definitive to be acceptable to 
him. Thus he would meet the criticism that the transaction version of the 
quantity equation has been relatively barren because it lumped various 
things together, the economic significance of which was quite diverse, by 
arguing for expansion of the formulation until each diverse item was 
treated separately. 

Whatever the ultimate appeal of such a program, it seems to me that, 
in the present state of economics and probably also in the foreseeable 
future, all workable relationships and analyses are bound to be both in- 
complete and "hybrid" in the sense that they summarize complicated 
variables. The important thing is that the relationship chosen should be 
"strategic" and, if possible, relatively constant, in order to eliminate im- 
mediate need for the more complete analysis which we are not yet in a 
position to undertake. In fact, the very reason that the quantity equation 
was originally so analytically useful and has continued as such an impor- 
tant teaching aid is that it summarized all the manifold forces influencing 
prices into exactly three variables. 17 Hence Marget's plea is not so much 
for what has been achieved by the transactions form of the quantity 
equation as it has been used, but for what might be achieved in the future 
if it were possible to break down the summary averages of the original 

16 V . cit., Vol. I, p. 538 ff. 

17 While progress in understanding our complex economic environment will undoubt- 
edly require increasingly complicated analysis, most recent progress has taken the form of 
developing more "strategic" groupings of a quite small number of variables. This is true 
not only of the shift from the transactions to the income version of the quantity equation, 
but also of the evolution of Keynes' ideas from the Treatise to the General Theory. 


equation into all the components necessary to deal with all separable 
price levels. One can, I believe, with all sincerity wish such an under- 
taking well and at the same time doubt both its probable achievements 
and the impact that it is likely to have on monetary theory in the imme- 
diate future. 

The importance of not only monetary theory but also monetary policy 
declined in the latter half of the logo's. For once a "reflation" such as 
was achieved by 1935 has been brought about, further increases in the 
money supply by Central Bank action alone are likely to lead to broadly 
compensating decreases in income velocity— at least over the range of in- 
creases which are possible without arousing insuperable political opposi- 
tion; hence at such times monetary control devices are not likely to be of 
much aid in combating a depression. From the point of view of control, 
therefore, their main use would have been to prevent a boom from getting 
out of hand after the existing supply of excess or idle balances had been 
exhausted. Our failure to recover fully before the outbreak of the war 
meant that they did not have a chance to undertake this modest role, and 
now even this role is likely to have been reduced, if not largely elimi- 
nated, as a result of the repercussions which the use of such controls 
would entail on the debt structure which we have inherited from the 
war. As a result we face an urgent need at the present time to develop 
alternative methods of control; this problem will be more fully considered 
when our postwar heritage is discussed below. 

III. Monetary Equilibrium, Period Analysis, and the 

General Theory 

As the extent to which the banking system could vary the money 
supply became clear, efforts had been made on the Continent, and espe- 
cially in the Swedish literature, to formulate what would have occurred 
"naturally" or "normally" in the absence of monetary "disturbances." 
Keynes' Treatise was primarily responsible for drawing attention, in 
Great Britain and the United States, to the resulting concept of an "equi- 
librium" in which money would be "neutral" in its effects on the econ- 
omy. Changes in money were thought of as being brought about by the 
rate of interest, which was either so low as to cause banks to create addi- 
tional funds to be added to those in existence or so high as to induce 
people to pay off bank loans and in this way reduce the money supply. 
Wicksell's pioneer formulation ran in terms of discrepancies between the 
"market" or actual rate of interest and the "natural" rate, which he de- 
fined as that rate which would keep prices constant, as he was working at 


a time when the major emphasis was on price stability. 18 This meant that 
an increase in the money supply equal to the increase not only in popu- 
lation but also in productivity would be "natural" under the definition 
used. On the other hand, Hayek, following the Austrian tradition of 
studying the effects of the imposition of money on a completely ' non- 
monetary" economy, defined the "natural" rate as that which would keep 
the effective quantity of money (money times velocity) constant, so that 
it would be "natural" for the price level of output to fall during periods 
of technological progress or when the supply of the productive factors 
was increasing. In other words, Hayek's definition meant that when there 
was no divergence between the two rates of interest, the level of the na- 
tional money income would be constant. Finally, Keynes in the Treatise, 
as we have seen, defined "normal" income as that which provided just 
enough entrepreneurial income to maintain the present level of employ- 
ment and output at present factor prices. 

The fundamental common problem faced by all these analyses was to 
define "equilibrium"; in the main it was the difficulty of giving meaning 
to this concept that caused the whole approach to be abandoned. This is 
true even when the emotional connotations of "natural," "normal," or 
"neutral" are discarded and the problem is stated in terms of "equilib- 
rium" without normative significance; and it also applies whether the 
mechanism of change is stated in terms of discrepancies between saving 
and investment or "market" and "natural" rates of interest. 

For example, as the role of payment practices and the degree of busi- 
ness integration in determining income velocity became clearer, even a 
theoretical definition of a "natural" rate which would eliminate "mone- 
tary" disturbances when there were changes in these factors became in- 
creasingly difficult. 19 For the distinction between "real" and "monetary" 
factors is a tenuous one at best. It was often argued that a release of 
money as a result of the reduction in money payments that follows from 
increased business integration should be offset if money was to remain 
"neutral"; but would not integration reduce exchange value even in a 
barter economy and therefore constitute a "real" rather than a "mone- 
tary" factor? Or should the release of money resulting from integration 
be offset only to the extent that it exceeded the decrease of exchange 
value that would have taken place in a barter regime? 

It was the Treatise more than any other volume that brought the prob- 
lem of equilibrium to a head and represented a crossroads in the develop- 

18 Constant prices in this formulation also made saving equal to investment. For a sum- 
mary of some of the other meanings given the "natural" rate by Wicksell, see A. W. 
Marget, of. cit., Vol. I, pp. 201-204. 

19 Cf. G. Haherler, Prosperity and Degression, 3rd ed. (Geneva, 1941), pp. 61-62. 


ment of monetary theory. 20 For the fundamental distinction that Keynes 
made in the Treatise between "normal" income and "profits" premised 
the existence of a lag in the adjustment of factor contracts, entrepreneur- 
ial commitments, or both: if entrepreneurs revised their commitments or 
labor reopened its contracts just as soon as there was a change in income, 
then there could be no difference between "normal" income and "profits." 
In the Treatise Keynes showed little interest in this problem, except to 
argue that a sufficient lag did exist to make his distinction between actual 
and "normal" income worth while. As a result the Treatise was criticized 
both because of the ambiguity of its discussion of the lag involved and for 
the extent to which the time period in question could be expected to 
vary over the course of the business cycle. Against these criticisms there 
were two possible lines of defense: either the various factors influencing 
the revision of contracts could be examined, and explicit assumptions 
could be made regarding the time period in question; or a "timeless" 
analysis could be developed and the problem avoided in this fashion. In 
the first of these directions lies modern period analysis, with its explicit 
assumptions regarding lags and leads, the fixity of contracts, and similar 
factors; in the other, the instantaneous analysis of Keynes* General 
Theory of Employment, Interest and Money. 21 

It is too early to attempt any definitive appraisal of the relative fruit- 
fulness of the two approaches, but it seems fair to say that round one has 
gone to the instantaneous approach. I should make clear that in judging 
"fruitfulness" I am laying major weight on the impact on public policy 
that has been or seems likely to be achieved before the economic system 
under study changes so drastically as to move the whole matter into the 
field of the economic historian. For however much the careful step-by- 
step procedure of period analysis commends itself as the only way to 
attain complete knowledge of the operation of our economic system, to 
date most examples can best be described as methodological explorations 
rather than positive contributions. 22 The difficulties to be faced are for- 
midable. Least important, perhaps, is the criticism leveled against the 
Treatise to the effect that entrepreneurs are always out of adjustment 

20 Chronologically Hayek follows the Treatise, but his methodological approach really 
belongs with the analyses of the preceding period. 

21 London, 1936. 

22 The best example of D. H. Robertson's work is his article in the Economic Journal 
(September 1933, XLIII) and the best summary of the Stockholm School is that of Bertil 
Ohlin, also in the Economic Journal (March and June 1937, XL VII). A possible excep- 
tion to the generalization in the text and the outstanding example of sustained work 
along period analysis lines is J. R. Hicks' Value and Capital (Oxford, 1939), although, 
as the title suggested, the author is not mainly concerned with problems in the field of 
monetary and business cycle theory. 


during a period of expansion or contraction. 23 Once the emphasis shifts 
away from "equilibrium" to period analysis, it becomes clear that lack of 
adjustment is to be expected. For it is the purpose of such analysis to 
show why the economy is out of balance, what is done about it, and what 
the consequences are. 

Far more important is the fact that no satisfactory bridge has been built 
between a mechanical analysis in which income received in one period is 
disposed in the next and an expectational analysis in which emphasis is 
placed on the extent to which the expectations held at the start of the 
period are in fact realized during the period. The mechanical approach 
puts major emphasis on such things as the flow of funds through the 
economic system from producer to income recipient and back again and 
the expansion or contraction of output through successive intervals of 
time. It has the advantage of showing how various changes take place 
within the institutional framework of the particular economy; but, be- 
cause it does not deal with expectations, it gives little light on many of 
the factors responsible for the changes involved. The expectational ap- 
proach, on the other hand, just because it does not demonstrate in step- 
by-step fashion the way in which funds move through an economy or 
output changes, often finds itself dealing with expectations which are 
inevitably doomed to disappointment from the start, as they involve a 
change of output or a movement of funds faster than the institutional 
arrangements of the system permit. 24 As with Professor Marget, who in 
fact advocates a form of period analysis, one can wish period analysis 
every success and yet remain skeptical as to whether it will prove fruitful 
within the immediate future even for problems which the Keynesian ap- 
proach has been least successful in handling. 25 

In contrast to the complexities of period analysis, the approach of the 
General Theory attempts to explain changes in the level of economic 
activity by means of a handful of variables: the quantity of money and 
liquidity f reference determine the rate of interest; the rate of interest and 

23 R. F. Harrod, The Trade Cycle (Oxford, 1936), p. 66. 

" 4 Hicks' analysis (op. cit.^) again comes closest to meeting this problem, tut the 
degree of simplification involved seems to me larger than can ultimately be accepted for 
monetary and business cycle theory. For income recipients at least, Robertson's analysi? 
falls mainly in the first class, while most of the Swedish work falls in the second, al- 
though the line between the two approaches is not always sharp. 

25 Haberler in the course of an extended discussion of period analysis (op. cit., pp. 177- 
195) suggests that the mechanical and expectational approaches are likely to come to- 
gether because the concept of expectations regarding uses of future income raises so many 
difficulties that the time period at issue may be shortened until the expectations are 
related to income actually realized in some past period— in other words, to Robertson's 
''disposable" income. If this in fact is to be the bridge between the two approaches, then 
the doubts expressed regarding fruitfulness seem confirmed, because so short a period 
would eliminate much of the content of the expectational approach. 


the marginal efficiency of capital determine the level of investment; and 
the level of investment and the marginal propensity to consume deter- 
mine income, output, and employment. In his enthusiasm for explaining 
"dynamic" changes in the simplest possible terms Keynes is sometimes 
reminiscent of the quantity theorist in believing that his analysis explains 
rather more than it actually does. In the marginal propensity to consume 
and the multiplier, for example, Keynes wanted and thought he had a 
largely constant factor which would permit us to say that "when there is 
an increment of aggregate investment, income will increase by an amount 
which is [the multiplier] times the increment of investment" 26 — just as a 
quantity theorist would say that, when there is an increase in money, 
income will increase by an amount which is income velocity times the 
increase in money. 

If it were true that the propensity to consume were relatively constant, 
Keynes would of course be correct in taking investment as the major 
variable, just as the quantity theorist took money when changes in in- 
come velocity were small. In the General Theory considerable space was 
devoted to arguing that the propensity reflected a stable psychological 
law which applied over wide ranges of income and broad periods of 
time. 27 But the stability of the propensity has been widely questioned in 
theoretical discussion. 28 Moreover, the statistical attempts to verify the 
stability of the consumption function have run into serious difficulties; 
the main source of error in the predictions of postwar income and em- 
ployment made toward the end of the war was apparently the result of 
inadequate estimates of the possible level of consumption, which in turn 
appear to have been caused by overestimates of the stability of the con- 
sumption function. 29 

26 General Theory, p. 115. My italics. The multiplier is equal to one divided by one 
minus the propensity to consume. 

27 The notation of Keynes reflects his belief in the constancy of the "marginal" pro- 
pensity to consume by making it equal to what I would think should be described as the 

1 dC 
"average" propensity to consume. Thus he writes A Y w = fc A I w , where 1 — t = - = w 

(General Theory, p. 115); but this can only be true if y W - = , v w . Hence if"Ty — rep- 

A.* w #i w w . 

resents the "average" propensity to consume for the change in income AY W , Keynes is in 
fact assuming that the "marginal" and "average" propensity are the same— or alterna- 
tively that the "marginal" propensity is constant— over the range of income AY W . 

28 B. Ohlin, "Some Notes on the Stockholm Theory of Savings and Investment," 
Economic Journal, June 1937, XL VII, pp. 221-240; D. Robertson, "A Survey of Modern 
Monetary Controversy," The Manchester School, 1938, pp. 133-153; and G. Haberler, 
op. cit., pp. 222-232, are among the critics of the alleged stability. For further references, 
see below, p. 329, note 30. 

29 See W. Woytinsky, "What Was Wrong in Forecasts of Postwar Depression," 
Journal of Political Economy, April 1947, LV, pp. 1 42-1 51, and references there cited. 


Certainly much of the appeal of the Keynesian approach lay in the 
stability of the multiplier which Keynes premised. For the idea that, once 
investment was given, saving, income, and employment would all fall 
into line through the operation of a (more or less) uniquely determined 
multiplier gave a certain grandeur to the analysis, which made it appear 
capable of explaining a wide variety of situations and therefore quite 
"dynamic" in character. When it is realized that the marginal multiplier 
(for small changes in investment), the average multiplier (for appreci- 
able changes), and the total multiplier (for investment as a whole) may 
have substantially different values, the analysis comes to be seen as con- 
siderably more limited and pedestrian in its scope and therefore more 
"static" in character. But the set of relationships which Keynes set forth- 
even if some of the components are less constant than he cared to admit 
—will certainly have a continuing impact on economic thinking because 
the variables he related to one another are of fundamental importance to 
any understanding of the problems with which he was concerned. 

Any more complete attempt to appraise the full impact of the Keynes- 
ian approach, even if it were as yet possible to do so, would take us 
beyond the confines of this paper. But it is perhaps worth while to con- 
clude by pointing out that all that has been said regarding the alternative 
approaches can be rephrased in terms of the saving-investment contro- 
versy. For it was the failure to keep factor contracts and entrepreneurial 
commitments up to date which was responsible for the difference in the 
Treatise between actual and "normal" income and therefore between 
saving and investment. Saving was confined to the income involved in 
the contracts and commitments made at the start of the "period," while 
investment was related to the income actually realized at the end of the 
"period," which could, of course, be more or less than that involved at 
the start. As a result a major factor determining the size of the discrep- 
ancy between saving and investment was the speed with which contracts 
and commitments were revised— the slower the revision the larger the dis- 
crepancy. Had the "period analysis" character of the difference between 
saving and investment in the Treatise been more fully recognized, it is 
possible that the advent of the General Theory would not have been 
marked by the extended and largely fruitless controversy as to whether 
saving and investment are equal or unequal. 30 

What the General Theory did in effect was to stress that during any 
period saving was equal to spending on investment. (This follows be- 
cause saving was defined as income less consumption, and income is 

30 Sixteen of the major articles on this subject are cited in H. H. Villard, op. cit., p. 28, 


equal to spending on total output and consumption to spending on con- 
sumption; hence by subtraction saving is equal to spending on invest- 
ment.) The main reason that this caused so much difficulty was that 
most economists have instinctively thought as consumers, who received 
income in the present period and then elected whether or not to spend it 
in a future period. 31 Hence most economists have typically— and fre- 
quently unconsciously— meant by saving the difference between the in- 
come of the present period and the consumption of a future period— a 
difference which might either be held idle or invested in the future 
period. In contrast, Keynes emphasized relationships within a single 
period, stressing chat the income of any given period would not have 
been received unless an identical amount of spending on consumption 
and investment had taken place. 32 What caused so much misunder- 
standing and difficulty was the mental adjustment involved in not pur- 
suing the usual more or less instinctive time sequence but instead identi- 
fying saving with the simultaneous spending on investment which gave 
rise to the income of the present period, rather than with the spending 
which might or might not take place in a future period. 

What Keynes succeeded in doing was to make clear that discrepancies 
between saving and investment, at least in the ex post meanings given 
the terms before the General Theory, depended on implicit or explicit 
period analysis. For when saving was thought of as income which was 
"hoarded" rather than spent on consumption, what must have been 
referred to was income of a period different from that in which the 
"hoarding" was thought of as taking place; for if the money involved 
has been "hoarded" in the sense of not being spent on output in the 
present period, then it would not have been part of present income. 33 
That it was desirable to make clear the "period analysis" character of all 
ex post differences between saving and investment is obvious. Yet in 
appraising the over-all effect of the way in which this matter was pre- 

31 The definition given by Keynes added to the confusion. For he defined saving as 
"the excess of income over consumption" (General Theory, p. 62). While actually the 
"excess" in any period— for the economy as a whole but not necessarily for every indi- 
vidual within the economy— is always identical with spending on investment during the 
period, the casual reader is likely to think of it as a sum which could be "hoarded." 

32 This does not deny that what happened in one period may influence what happens 
in a subsequent period, nor does it imply that the amount spent during a given period 
must "come out of" the income of that period, as this depends on the length of the 
period. If the period is so short that there is no time for any money to be spent more 
than once, then all the spending of the period will "come out of" previously unused 
cash balances and total income will be less than cash balances; if, however, the period 
is long enough, money may be spent often enough for total income to be a multiple of 
average cash balances. 

33 This is, of course, true only for the economy as a whole, as the "hoarding" of one 
individual may be offset by the "dishoarding" of another. 


sented in the General Theory, I think it fair to say that it greatly impeded 
progress in economic thinking— and this despite the fact that the General 
Theory as a whole certainly made the greatest contribution to our ulti- 
mate understanding of economic fluctuations of any volume published 
in the decade of the 'thirties. For it was a paradox of Keynes' greatness 
that he treated what was a minor clarification of concept as a great new 
discovery, thereby completely confounding his less nimble-witted col- 
leagues—though it is only fair to admit that Keynes' disciples were fre- 
quently fins royalistes que le rot. The resulting years of controversy were 
only ended by the war; their effect was not only to divert much effort 
of economists into "translating" Keynes into more conventional terms 
but also to present to the layman the spectacle of a science deeply divided. 
It is perhaps the ultimate irony of his career that Keynes, with his intense 
interest in practical programs to reduce business fluctuations, should have 
contributed so much to the failure of American economists as a group 
either to develop an agreed program for mitigating the business cycle or 
to carry any appreciable weight in public decisions on matters of economic 

IV. Liquidity Preference and Interest 

During recent years Keynes' General Theory represents the outstand- 
ing development in interest theory, so that it is appropriate to start with 
a consideration of that volume. It is the contention of the General Theory 
that the rate of interest is entirely determined by two factors: the supply 
of money and liquidity preference; in other words, liquidity preference 
is a function which relates the demand for money to the rate of interest. 
Keynes argues that people have three reasons for desiring "liquidity": 
the transactions-motive, the precautionary-motive, and the speculative- 
motive. The first of these is the familiar concept of balances needed to 
bridge the gap, for both business and income-recipients, between receipts 
and expenditures connected with current output; 34 the second is "to 
provide for contingencies requiring sudden expenditure and for unfore- 
seen opportunities of advantageous purchases" and is thought of as vary- 
ing with the level of income; 35 and the third is to secure "profit from 

34 To the more usual formulation Keynes added the need for funds "due to the time- 
lag between the inception and the execution of the entrepreneurs' decisions," which he 
called the demand for "finance." See "Mr. Keynes and Finance: Comment," Economic 
Journal, June 1938, XLVIII, p. 319. The fundamental structure of the Keynesian analy- 
sis is unaffected by this addition, which is simply another factor adding to the demand 
for transaction (and probably also precautionary) balances. 

35 General Theory, p. 196. The distinction between precautionary and speculative 
balances has always seemed to me finely drawn. 


knowing better than the market what the future will bring forth'— in 
other words, from the expectation that money will decline in value less 
than other assets. 36 In short, as Keynes uses them, transaction and pre- 
cautionary balances are "active" balances held in connection with the 
production of current income and speculative balances are "idle" bal- 
ances held on capital account. Note that it is all these balances which 
are related to the interest rate by liquidity preference; hence "liquidity 
preference," as Keynes uses the term, covers considerably more than 
a speculative desire to hold assets in liquid form because it is thought 
that illiquid assets are likely to depreciate in value. 37 

As with so much of Keynes' work, an appreciable part of the novelty 
of his treatment of interest arises from either terminological innovations 
or unusual assumptions. Take, for example, the fact that in Keynes' 
formulation changes in the desire to save appear not to have any effect 
on the interest rate. Keynes tells us that economists have generally 
assumed "that, ceteris paribus, a decrease in spending will tend to 
lower the rate of interest and an increase in investment to raise it. But 
if what these two quantities determine is, not the rate of interest, but the 
aggregate volume of employment, then our outlook on the mechanism 
of the economic system will be profoundly changed. A decreased readi- 
ness to spend will be looked on in a quite different light if, instead of 
being regarded as a factor which will, ceteris paribus, increase investment, 
it is seen as a factor which will, ceteris paribus, diminish employment." 38 
To what extent is this a real and not merely an apparent contrast with 
the usual formulation, in which changes in the "readiness to spend"— 
or in saving in a non-Keynesian sense— are thought of as having an 
important influence on the rate of interest? 

Actually Keynes' startling conclusion that "a decreased readiness to 
spend" will diminish employment rather than increase investment follows 
directly from the fact that he includes liquidity preference within the 
ceteris paribus assumption; in other words, he assumes that liquidity 
preference is unaltered despite a "decreased readiness to spend." But 
this is another way of saying that the individual wishes to hold idle the 
money he was previously ready to spend; for if the quantity of money 
and liquidity preference (and therefore the rate of interest) are un- 
changed, then a decrease in spending can only mean that the funds 
involved have been shifted from transaction and precautionary balances 

SB lbid., p. 170. 

37 Haberler suggests that the relationship between speculative balances and the rate 
of interest be called "liquidity preference proper" to distinguish it from the relationship 
between all balances and the rate of interest (op. cit., p. 210). 

38 General Theory, p. 185. 


to speculative balances. 39 Under these circumstances investment need 
not increase and employment as a result will fall. But there is no reason 
why liquidity preference must remain unchanged, and when it is re* 
moved from ceteris paribus, quite different results from those which 
Keynes indicates are possible. For the money freed by the "decreased 
readiness to spend" may well decrease the individual's liquidity prefer- 
ence, which in turn can be expected to reduce the rate of interest and 
increase investment, exactly as in the more conventional formulations. 40 
Had Keynes said that when an individual saves in order to "hoard," the 
social effects are quite different than when an individual saves in order 
to invest, his meaning would have been clearer but his statement less 

Of course, the concept of "hoarding" is not a part of the Keynesian 
system. This is understandable because the instantaneous approach of 
the General Theory avoids so far as possible specific reference to time 
periods, while "hoarding" in its usual meaning must have a time dimen- 
sion. For "hoarding" which is timeless becomes identical with holding 
money; accordingly, as all money must be held by someone at all times 
if it is to be counted as money, it becomes correct to say that all money 
is "hoarded" and that changes in "hoarding" from one period to the 
next are the same thing as changes in the quantity of money. From this 
it follows that "it is impossible for the actual amount of hoarding to 
change as a result of decisions on the part of the public, so long as we 
mean by 'hoarding* the actual holding of cash. For the amount of hoard- 
ing must be equal to the quantity of money . . . ; and the quantity of 
money is not determined by the public." 41 Here again is a startling 
result based on an unusual meaning for a common term; but in this case 
the usage on which the result depended was reasonably clear. 

What have these changes and innovations contributed to interest 
theory? The pervading emphasis which Keynes has laid on the depend- 
ence of saving and interest on the level of income has been of great im- 
portance. The "classical" theory of saving and interest had been most 
concerned with long-run problems in which it seemed appropriate to take 
the level of income as more or less fixed and to investigate the forces 
determining the amount of such income which would be saved and 

39 Following Haberler's suggestion, the situation is one in which "liquidity preference 
proper" has increased sufficiently to absorb the money freed by the "decreased readiness 
to spend." 

40 Again following Haberler, if there is no change in the individual's "liquidity prefer- 
ence proper," the money freed by the "decreased desire to spend" can be expected to act 
on the rate of interest and the level of investment in the same way as any decrease ip 
over-all liquidity preference. 

^General Theory, p. 174. 


invested. Keynes was by no means the first person to indicate that saving 
and interest were influenced by the level of income and much of his 
criticism of ' classical" theory, if it was meant to apply to all the work of 
all his predecessors and not to those "real capital" theorists who were 
primarily concerned with long-run equilibrium, can only be characterized 
as overly exuberant. In fact, Keynes himself came to agree that he was 
"shying at a composite Aunt Sally of uncertain age." 42 But exuberance 
aside, Keynes clearly deserves credit for emphasizing the extent to 
which an increase in investment, working through an increase in income, 
could be expected to provide an offsetting quantity of saving. In part 
this emphasis was the result of the definitional identity between saving 
and investment; but back of this lay the real fact that large changes in 
saving and investment were possible with little change in the level of 
interest if accompanied by large changes in income. In fact it is quite 
possible that the start of an upturn will bring such a release of specu- 
lative (idle) balances that at least the early periods of recovery may be 
accompanied by a lower rate of interest than that which had previously 

In addition to his emphasis on changes in income, Keynes' most im- 
portant contribution has been the insight which he has given us on the 
behavior of speculative balances, both in general and especially as a 
result of changing anticipations regarding the rate of interest. Applied to 
perpetual bonds, which present the simplest as well as the most extreme 
case, Keynes points out that it is impossible for the rate anticipated a 
year hence to exceed the current rate by more than the square of the 
current rate. 43 For otherwise it would be more profitable to hold money 
than bonds, as the reduction in the capital value of such securities during 
the year as a result of the rise in the interest rate would be greater than 
the sum received as interest. Of course most bonds are not perpetual, 
so that rate increases in excess of the square of the current rate can be 
anticipated without causing a complete shift into idle balances. But 
clearly whenever appreciable rate increases are anticipated the effective- 
ness of monetary policy is greatly reduced, and recovery is likely to be 
slow even if vigorous action is taken by the monetary authorities. For 
when the recession phase has come to an end and prices of securities are 
high and yields low as a result of a reflationary "cheap money" policy, a 
time may come when any further expansion of the money supply will 
flow overwhelmingly into idle (speculative) balances because investors 

42 "The 'Ex- Ante* Theory of the Rate of Interest," Economic Journal, December 1937, 
XLVII, p. 663. 

43 General Theory, p. 202. If the current rate is 3 per cent, the anticipated rate cannot 
be more than 3.09 per cent. 


generally believe that the present low level of the interest rate will not 
be maintained. 44 

In appraising the probable importance in actual practice of such a 
situation, Keynes himself has repudiated the extreme possibility that 
all additional funds will flow into idle balances, stating that while this 
"might become practically important in future, I know of no example 
of it hitherto. Indeed, owing to the unwillingness of most monetary 
authorities to deal boldly in debts of long term, there has not been 
much opportunity for a test/' 45 Yet it is by assuming implicitly or ex- 
plicitly what is in effect an "absolute liquidity preference" under which 
the demand for idle (speculative) balances is insatiable, that Keynes 
achieves his most striking differences from other theorists. In appraising 
his contribution one has again to weigh the real insight that he has given 
us against the confusion that has resulted from his perennial inclination 
to treat an unusual, and therefore startling, situation as if it applied 
generally— in short, to make a "gen